AI pricing
AI API price history
Launch prices and every list-price change we could confirm for 120 models from OpenAI, Anthropic, Google, xAI, Mistral, DeepSeek, 2021 to 2026. Each row links its source. Updated 2026-11-09.
- Priciest launch
- GPT-4.5 (OpenAI)
- $75 / $150
- Cheapest launch
- Ministral 3B (Mistral)
- $0.04 / $0.04
- Biggest cut
- o3 (OpenAI)
- −80% on Jun 10, 2025
- Price changes
- 25 cuts, 3 increases
- 28 in all
Launch price of every model, 2021–2026
Blended price per 1M tokens at launch. Hover or tap a dot for the model.
Biggest price cuts
| Model | Date | Before (in / out) | After (in / out) | Change |
|---|---|---|---|---|
| o3 OpenAI | Jun 10, 2025 | $10 / $40 | $2 / $8 | −80% |
| GPT-5.6 Luna OpenAI | Jul 30, 2026 | $1 / $6 | $0.2 / $1.2 | −80% |
| Codestral Mistral | Sep 17, 2024 | $1 / $3 | $0.2 / $0.6 | −80% |
| Mistral Small Mistral | Sep 17, 2024 | $1 / $3 | $0.2 / $0.6 | −80% |
| Gemini 1.5 Flash Google | Aug 12, 2024 | $0.35 / $1.05 | $0.075 / $0.3 | −75% |
| DeepSeek V4 Pro DeepSeek | Apr 26, 2026 | $1.74 / $3.48 | $0.43 / $0.87 | −75% |
| GPT-3 davinci OpenAI | Sep 1, 2022 | $60 / $60 | $20 / $20 | −67% |
| Gemini 1.5 Pro Google | Oct 1, 2024 | $3.5 / $10.5 | $1.25 / $5 | −58% |
| Mistral Large Mistral | May 5, 2024 | $8 / $24 | $4 / $12 | −50% |
| Mistral Small Mistral | May 5, 2024 | $2 / $6 | $1 / $3 | −50% |
Every price event
USD per 1M tokens (input / output), standard tier, newest first. Download CSV.
2026
| 11-05 | Mistral Large 4 Mistral GA (mistral-large-4). Cached input $0.07. Mistral Large 3 deprecated 2026-10-31, retires 2027-04-30. | Price cut−49% | $0.7 / $2.1 | source |
| 10-26 | GPT-6.1 Terra OpenAI Launch list price. | Launch | $1 / $5 | source |
| 10-23 | o4-mini OpenAI Shut down in the April 22, 2026 legacy snapshot deprecation. | Retired | source | |
| 10-23 | o3-mini OpenAI Shut down in the April 22, 2026 legacy snapshot deprecation. | Retired | source | |
| 10-23 | o1 OpenAI Shut down in the April 22, 2026 legacy snapshot deprecation. | Retired | source | |
| 10-23 | GPT-4.1 nano OpenAI Shut down in the April 22, 2026 legacy snapshot deprecation. | Retired | source | |
| 10-23 | GPT-4 Turbo OpenAI Shut down in the April 22, 2026 legacy snapshot deprecation. | Retired | source | |
| 10-23 | GPT-4 OpenAI Shut down in the April 22, 2026 legacy snapshot deprecation. | Retired | source | |
| 10-23 | GPT-3.5 Turbo OpenAI Shut down in the April 22, 2026 legacy snapshot deprecation. | Retired | source | |
| 10-13 | Gemini 4 Argon Google Introductory price through Nov 30, 2026; $4.00 / $20.00 from Dec 1, 2026. Paid tier only. | Launch | $2 / $10 | source |
| 10-07 | Claude Haiku 5.5 Anthropic 90% below Haiku 4.5 ($1 / $5) for prompts up to 100K tokens. Prompts over 100K cost $0.50 / $2.50 for the whole request. | Launch | $0.1 / $0.5 | source |
| 10-06 | Mistral Large 4 Mistral Public preview. Docs showed this as the original price with a preview sale price of $0.68/$2.09 (2026-10-19 snapshot). | Launch | $1.36 / $4.18 | source |
| 09-29 | GPT-6.1 Sol OpenAI Same list price as GPT-6 Sol, one-fifth of GPT-6 Astra. | Launch | $2 / $10 | source |
| 09-28 | Claude Sonnet 5.5 Anthropic Same price as Sonnet 5. | Launch | $2 / $10 | source |
| 09-22 | Claude Opus 5.5 Anthropic 20% below Opus 5 ($5 / $25). Cache reads $0.20 per 1M. | Launch | $4 / $20 | source |
| 09-22 | GPT-6 Luna OpenAI Below GPT-5.6 Luna ($0.20 / $1.20). | Launch | $0.1 / $0.5 | source |
| 09-22 | GPT-6 Sol OpenAI 50% below GPT-5.6 Sol's promotional price. | Launch | $2 / $10 | source |
| 09-21 | Grok 4.7 xAI Under 200K prompt tokens; $4/$12 above. 500K context. | Launch | $2 / $6 | source |
| 09-10 | DeepSeek V4.1 Flash DeepSeek Replaces V4-Flash at a lower price (peak shown; off-peak $0.15/$0.60), effective 04:00 UTC. API name deepseek-flash. | Launch | $0.3 / $1.2 | source |
| 09-10 | DeepSeek-V4-Flash DeepSeek Retired; deepseek-v4-flash requests routed to V4.1-Flash at Flash price. | Retired | source | |
| 09-03 | GPT-6 Astra OpenAI Flagship; base tier, prompts up to 272K tokens. | Launch | $10 / $50 | source |
| 09-02 | Gemini 3.8 Flash Google Introductory price through Dec 31, 2026; $1.50 / $7.50 from Jan 1, 2027. | Launch | $0.75 / $3.75 | source |
| 09-01 | Claude Mythos 5.1 Anthropic Launched with Fable 5.1 at the same price. | Launch | $10 / $50 | source |
| 09-01 | Claude Fable 5.1 Anthropic Same price as Fable 5; cache reads cut 75% ($1 to $0.25 per 1M). | Launch | $10 / $50 | source |
| 08-31 | Mistral Medium 3 Mistral mistral-medium-2505 and 2508 (3.1); scheduled retirement date in Mistral docs. | Retired | source | |
| 08-21 | GPT-5.6 Sol OpenAI Promotional price: input -20%, output -33%. OpenAI says it runs until Nov 30, 2026. | Price cut−29% | $4 / $20 | source |
| 08-16 | DeepSeek V4 Pro DeepSeek Peak/off-peak pricing from 16:00 UTC. Peak shown; off-peak is half ($0.66/$1.98). Peak hours 01:00-04:00 and 06:00-10:00 UTC weekdays. | Price increase+264% | $1.32 / $3.96 | source |
| 08-16 | DeepSeek-V4-Flash DeepSeek Peak/off-peak pricing from 16:00 UTC. Peak shown; off-peak is half ($0.22/$0.66). | Price increase+277% | $0.44 / $1.32 | source |
| 08-13 | Gemini 3.6 Flash Google Same introductory price as 3.7 Flash, through Dec 31, 2026; back to $1.50 / $7.50 from Jan 1, 2027. The Jul 31, 2026 capture still showed $1.50 / $7.50. | Price cut−50% | $0.75 / $3.75 | source |
| 08-13 | Gemini 3.7 Flash Google Introductory price through Dec 31, 2026; $1.50 / $7.50 from Jan 1, 2027. | Launch | $0.75 / $3.75 | source |
| 08-12 | Grok 4.6 xAI Under 200K prompt tokens; $4/$12 above. 500K context. | Launch | $2 / $6 | source |
| 08-05 | Claude Opus 4.1 Anthropic claude-opus-4-1-20250805 retired (deprecated 2026-06-05). | Retired | source | |
| 07-31 | Mistral Small 3.2 Mistral mistral-small-2506; scheduled retirement date in Mistral docs. | Retired | source | |
| 07-30 | GPT-5.6 Terra OpenAI Cut 20% (announced in changelog Jul 30, 2026); new price listed on the model page. | Price cut−20% | $2 / $12 | source |
| 07-30 | GPT-5.6 Luna OpenAI Cut 80% (announced in changelog Jul 30, 2026); new price listed on the model page. | Price cut−80% | $0.2 / $1.2 | source |
| 07-24 | Claude Opus 5 Anthropic Same price as Opus 4.8. | Launch | $5 / $25 | source |
| 07-21 | Gemini 3.5 Flash-Lite Google GA. Same input price for text, image, video and audio. | Launch | $0.3 / $2.5 | source |
| 07-21 | Gemini 3.6 Flash Google GA. Output cheaper than 3.5 Flash ($9.00); input unchanged. | Launch | $1.5 / $7.5 | source |
| 07-16 | Grok 4.5 xAI | Launch | $2 / $6 | source |
| 07-09 | GPT-5.6 Luna OpenAI Cheapest tier. | Launch | $1 / $6 | source |
| 07-09 | GPT-5.6 Terra OpenAI Balanced tier. | Launch | $2.5 / $15 | source |
| 07-09 | GPT-5.6 Sol OpenAI Flagship of the GPT-5.6 family; base tier, prompts up to 272K tokens. | Launch | $5 / $30 | source |
| 06-30 | Claude Sonnet 5 Anthropic Launched at an introductory $2 / $10, with $3 / $15 set to apply from 2026-09-01. | Launch | $2 / $10 | source |
| 06-15 | Claude Sonnet 4 Anthropic claude-sonnet-4-20250514 retired (deprecated 2026-04-14). | Retired | source | |
| 06-15 | Claude Opus 4 Anthropic claude-opus-4-20250514 retired (deprecated 2026-04-14). | Retired | source | |
| 06-09 | Claude Mythos 5 Anthropic Launched with Fable 5 at the same price. | Launch | $10 / $50 | source |
| 06-09 | Claude Fable 5 Anthropic New top tier above Opus. 1M context at standard price. | Launch | $10 / $50 | source |
| 06-01 | Gemini 2.0 Flash-Lite Google Shut down (gemini-2.0-flash-lite and gemini-2.0-flash-lite-001). | Retired | source | |
| 06-01 | Gemini 2.0 Flash Google Shut down (gemini-2.0-flash and gemini-2.0-flash-001). | Retired | source | |
| 05-31 | Mistral Large 2.1 Mistral mistral-large-2411; scheduled retirement date in Mistral docs. | Retired | source | |
| 05-29 | Grok Build 0.1 xAI Coding model, public beta on API. | Launch | $1 / $2 | source |
| 05-28 | Claude Opus 4.8 Anthropic Same price as Opus 4.7. Fast mode costs $10 / $50. | Launch | $5 / $25 | source |
| 05-19 | Gemini 3.5 Flash Google GA at Google I/O 2026. One price at all context lengths. | Launch | $1.5 / $9 | source |
| 05-15 | grok-code-fast-1 xAI Retired 12:00 PT; requests redirect to grok-4.3 (grok-code-fast-1 to grok-build-0.1) and bill at the new model's price. | Retired | source | |
| 05-15 | Grok 4.1 Fast xAI Retired 12:00 PT; requests redirect to grok-4.3 (grok-code-fast-1 to grok-build-0.1) and bill at the new model's price. | Retired | source | |
| 05-15 | Grok 4 Fast xAI Retired 12:00 PT; requests redirect to grok-4.3 (grok-code-fast-1 to grok-build-0.1) and bill at the new model's price. | Retired | source | |
| 05-15 | Grok 4 xAI Retired 12:00 PT; requests redirect to grok-4.3 (grok-code-fast-1 to grok-build-0.1) and bill at the new model's price. | Retired | source | |
| 05-15 | Grok 3 xAI Retired 12:00 PT; requests redirect to grok-4.3 (grok-code-fast-1 to grok-build-0.1) and bill at the new model's price. | Retired | source | |
| 05-06 | Grok 4.20 xAI Cut to Grok 4.3 price. Exact date unknown: $2/$6 on 2026-04-28 snapshot, $1.25/$2.50 on 2026-05-06 snapshot. | Price cut−48% | $1.25 / $2.5 | source |
| 04-30 | Grok 4.3 xAI Under 200K prompt tokens; double at or above 200K. 1M context. Launch date per VentureBeat (2026-05-01, 'last night'). | Launch | $1.25 / $2.5 | source |
| 04-28 | Mistral Medium 3.5 Mistral Up from $0.40/$2 for Medium 3/3.1. Deprecated 2026-10-31, retires 2027-04-30. | Launch | $1.5 / $7.5 | source |
| 04-26 | DeepSeek V4 Pro DeepSeek Limited-time 75% discount (first seen 2026-04-26 pricing page; DeepSeek posted it 2026-04-27). Extended to 2026-05-31, then made the permanent list price. | Price cut−75% | $0.43 / $0.87 | source |
| 04-24 | DeepSeek V4 Pro DeepSeek 1M context. Cache hit $0.145. | Launch | $1.74 / $3.48 | source |
| 04-24 | DeepSeek-V4-Flash DeepSeek 1M context. Cache hit $0.028 (cut to $0.0028 on 2026-04-26). | Launch | $0.14 / $0.28 | source |
| 04-24 | DeepSeek-V3.2 DeepSeek deepseek-chat / deepseek-reasoner now routed to V4-Flash; the two names were shut off 2026-07-24 15:59 UTC. | Retired | source | |
| 04-24 | GPT-5.5 OpenAI Released in the API Apr 24, 2026 (ChatGPT first). Base tier, prompts up to 272K tokens. | Launch | $5 / $30 | source |
| 04-20 | Claude 3 Haiku Anthropic claude-3-haiku-20240307 retired (deprecated 2026-02-19). | Retired | source | |
| 04-16 | Claude Opus 4.7 Anthropic Same price as Opus 4.6. New tokenizer produces about 30% more tokens for the same text. | Launch | $5 / $25 | source |
| 03-17 | GPT-5.4 nano OpenAI API only. | Launch | $0.2 / $1.25 | source |
| 03-17 | GPT-5.4 mini OpenAI Launch list price. | Launch | $0.75 / $4.5 | source |
| 03-16 | Mistral Small 4 Mistral Up from $0.10/$0.30 for Small 3.x. | Launch | $0.15 / $0.6 | source |
| 03-11 | Grok 4.20 xAI grok-4.20-beta-0309 (reasoning, non-reasoning, multi-agent). xAI release notes say March 2026; date is the earliest archived docs listing. | Launch | $2 / $6 | source |
| 03-09 | Gemini 3 Pro Google gemini-3-pro-preview shut down; the alias now points to gemini-3.1-pro-preview. | Retired | source | |
| 03-05 | GPT-5.4 OpenAI Base tier, prompts up to 272K tokens; longer prompts cost more. | Launch | $2.5 / $15 | source |
| 03-03 | Gemini 3.1 Flash-Lite Google Preview; text/image/video input, audio $0.50. Same price at GA (May 7, 2026). | Launch | $0.25 / $1.5 | source |
| 02-19 | Gemini 3.1 Pro Google Preview (gemini-3.1-pro-preview). Prompts up to 200K tokens; $4.00 / $18.00 above 200K. | Launch | $2 / $12 | source |
| 02-19 | Claude 3.5 Haiku Anthropic claude-3-5-haiku-20241022 retired (deprecated 2025-12-19). | Retired | source | |
| 02-19 | Claude 3.7 Sonnet Anthropic claude-3-7-sonnet-20250219 retired (deprecated 2025-10-28). | Retired | source | |
| 02-17 | Claude Sonnet 4.6 Anthropic Same price as Sonnet 4.5. Full 1M context at standard price. | Launch | $3 / $15 | source |
| 02-05 | Claude Opus 4.6 Anthropic Same price as Opus 4.5. Full 1M context at standard price. | Launch | $5 / $25 | source |
| 01-05 | Claude 3 Opus Anthropic claude-3-opus-20240229 retired (deprecated 2025-06-30). | Retired | source |
2025
| 12-31 | Ministral 8B Mistral Scheduled retirement date in Mistral docs; replaced by Ministral 3 8B. | Retired | source | |
| 12-31 | Ministral 3B Mistral Scheduled retirement date in Mistral docs; replaced by Ministral 3 3B. | Retired | source | |
| 12-17 | Gemini 3 Flash Google Preview (gemini-3-flash-preview). Text/image/video input; audio $1.00. | Launch | $0.5 / $3 | source |
| 12-11 | GPT-5.2 OpenAI Priced 40% above GPT-5.1. | Launch | $1.75 / $14 | source |
| 12-02 | Mistral Large 3 Mistral Open-weight; 75% below Mistral Large 2.1 ($2/$6). | Launch | $0.5 / $1.5 | source |
| 12-02 | Ministral 3 8B Mistral ministral-8b-2512; up from $0.10 for Ministral 8B. | Launch | $0.15 / $0.15 | source |
| 12-02 | Ministral 3 3B Mistral ministral-3b-2512; up from $0.04 for Ministral 3B. | Launch | $0.1 / $0.1 | source |
| 12-02 | Ministral 3 14B Mistral ministral-14b-2512. | Launch | $0.2 / $0.2 | source |
| 11-30 | Mistral Small 3 Mistral mistral-small-2501 and 2503 retired (also Small 24.09). | Retired | source | |
| 11-30 | Codestral 25.01 Mistral codestral-2501 retired; replaced by codestral-2508 at the same $0.30/$0.90. | Retired | source | |
| 11-24 | Claude Opus 4.5 Anthropic Opus price cut by two thirds vs Opus 4.1 ($15 / $75). | Launch | $5 / $25 | source |
| 11-19 | Grok 4.1 Fast xAI Reasoning and non-reasoning variants. Cached input $0.05. | Launch | $0.2 / $0.5 | source |
| 11-18 | Gemini 3 Pro Google Preview (gemini-3-pro-preview). Prompts up to 200K tokens; $4.00 / $18.00 above 200K. Not on the free tier. | Launch | $2 / $12 | source |
| 11-13 | GPT-5.1 OpenAI Launch post says pricing is the same as GPT-5 (Nov 13, 2025). The GPT-5.2 post's price table lists gpt-5.1 at $1.25 / $10. | Launch | $1.25 / $10 | source |
| 10-28 | Claude 3.5 Sonnet Anthropic Both 3.5 Sonnet snapshots retired (deprecated 2025-08-13). | Retired | source | |
| 10-27 | o1-mini OpenAI Removed from the API; replacement o4-mini. | Retired | source | |
| 10-15 | Claude Haiku 4.5 Anthropic Priced above Claude 3.5 Haiku ($0.80 / $4). | Launch | $1 / $5 | source |
| 09-29 | DeepSeek-V3.2-Exp DeepSeek Prices cut 50%+ from $0.56/$1.68, effective 10:00 UTC. Cache hit $0.028. V3.2 (2025-12-01) kept this price. | Price cut | $0.28 / $0.42 | source |
| 09-29 | Gemini 1.5 Flash-8B Google Shut down (changelog, Sep 29, 2025). | Retired | source | |
| 09-29 | Gemini 1.5 Flash Google Shut down (changelog, Sep 29, 2025). | Retired | source | |
| 09-29 | Gemini 1.5 Pro Google Shut down (changelog, Sep 29, 2025). | Retired | source | |
| 09-29 | Claude Sonnet 4.5 Anthropic Same price as Sonnet 4. | Launch | $3 / $15 | source |
| 09-19 | Grok 4 Fast xAI Under 128K tokens; $0.40/$1.00 at or above 128K. | Launch | $0.2 / $0.5 | source |
| 09-05 | DeepSeek-V3.1 DeepSeek V3.1 (released 2025-08-21) serves both deepseek-chat and deepseek-reasoner; new single price from 2025-09-05 16:00 UTC (chat up from $0.27/$1.10, reasoner from $0.55/$2.19). Off-peak discounts ended. | Launch | $0.56 / $1.68 | source |
| 08-28 | grok-code-fast-1 xAI Cached input $0.02. | Launch | $0.2 / $1.5 | source |
| 08-07 | GPT-5 nano OpenAI Launch list price. | Launch | $0.05 / $0.4 | source |
| 08-07 | GPT-5 mini OpenAI Launch list price. | Launch | $0.25 / $2 | source |
| 08-07 | GPT-5 OpenAI Launch list price. | Launch | $1.25 / $10 | source |
| 08-05 | Claude Opus 4.1 Anthropic Same price as Opus 4 (model ID claude-opus-4-1-20250805; price from the current pricing page's legacy table). | Launch | $15 / $75 | source |
| 07-28 | o1-preview OpenAI Removed from the API; replacement o3. | Retired | source | |
| 07-21 | Claude 3 Sonnet Anthropic claude-3-sonnet-20240229 retired (deprecated 2025-01-21). | Retired | source | |
| 07-21 | Claude 2 / Claude 2.1 Anthropic claude-2.0 and claude-2.1 retired (deprecated 2025-01-21). | Retired | source | |
| 07-14 | GPT-4.5 OpenAI gpt-4.5-preview removed from the API; replacement gpt-4.1. | Retired | source | |
| 07-09 | Grok 4 xAI grok-4-0709; announced 2025-07-09. Price from archived xAI docs (2025-09-11). | Launch | $3 / $15 | source |
| 06-17 | Gemini 2.5 Flash-Lite Google Preview; text/image/video input. Same price at GA (Jul 22, 2025). | Launch | $0.1 / $0.4 | source |
| 06-17 | Gemini 2.5 Flash Google Price changed at GA: input up from $0.15, output down from $3.50 (thinking) and up from $0.60 (non-thinking); thinking/non-thinking split removed. | Price increase−14% | $0.3 / $2.5 | source |
| 06-16 | Mistral Small Mistral mistral-small-2402 retired. | Retired | source | |
| 06-16 | Mistral Medium Mistral mistral-medium-2312 retired. | Retired | source | |
| 06-16 | Mistral Large Mistral mistral-large-2402 retired. | Retired | source | |
| 06-16 | Codestral Mistral codestral-2405 retired. | Retired | source | |
| 06-10 | o3 OpenAI Cut 80% on the day o3-pro launched (changelog Jun 10, 2025). Price from archived pricing page (Jul 2, 2025 snapshot). | Price cut−80% | $2 / $8 | source |
| 06-06 | GPT-4 32K OpenAI gpt-4-32k, -0613 and -0314 shut down. | Retired | source | |
| 05-22 | Claude Sonnet 4 Anthropic Same price as earlier Sonnets. A 1M-token context beta added later charged $6 / $22.50 for prompts over 200K tokens. | Launch | $3 / $15 | source |
| 05-22 | Claude Opus 4 Anthropic Same price as Claude 3 Opus. | Launch | $15 / $75 | source |
| 05-07 | Mistral Medium 3 Mistral Same price kept by Medium 3.1 (2025-08-12). | Launch | $0.4 / $2 | source |
| 04-17 | Gemini 2.5 Flash Google Preview. Output $3.50 with thinking, $0.60 without thinking. Audio input $1.00. | Launch | $0.15 / $3.5 | source |
| 04-16 | o4-mini OpenAI Price from archived pricing page (May 3, 2025 snapshot). | Launch | $1.1 / $4.4 | source |
| 04-16 | o3 OpenAI Price from archived pricing page (May 3, 2025 snapshot). | Launch | $10 / $40 | source |
| 04-14 | GPT-4.1 nano OpenAI OpenAI's first nano model. | Launch | $0.1 / $0.4 | source |
| 04-14 | GPT-4.1 mini OpenAI Launch list price. | Launch | $0.4 / $1.6 | source |
| 04-14 | GPT-4.1 OpenAI Launch list price. | Launch | $2 / $8 | source |
| 04-09 | Grok 3 mini xAI Launched alongside Grok 3; price from archived xAI docs (2025-04-18). | Launch | $0.3 / $0.5 | source |
| 04-09 | Grok 3 xAI API launch date per press reports; price from archived xAI docs (2025-04-18). grok-3-fast was $5/$25. | Launch | $3 / $15 | source |
| 04-04 | Gemini 2.5 Pro Google Paid preview (gemini-2.5-pro-preview-03-25). Prompts up to 200K tokens; $2.50 / $15.00 above 200K. Output includes thinking tokens. Same price at GA (Jun 17, 2025). | Launch | $1.25 / $10 | source |
| 03-30 | Mistral Large 2 Mistral mistral-large-2407 retired. | Retired | source | |
| 02-27 | GPT-4.5 OpenAI gpt-4.5-preview, research preview. Price from archived pricing page (Mar 3, 2025 snapshot). | Launch | $75 / $150 | source |
| 02-26 | DeepSeek-V3 DeepSeek Off-peak discount added: 50% off 16:30-00:30 UTC daily (off-peak $0.135/$0.55). Standard price unchanged at $0.27/$1.10. | Price cut | source | |
| 02-26 | DeepSeek-R1 DeepSeek Off-peak discount added: 75% off 16:30-00:30 UTC daily (off-peak $0.135/$0.55). Standard price unchanged at $0.55/$2.19. | Price cut | source | |
| 02-25 | Gemini 2.0 Flash-Lite Google GA release; the preview (Feb 5, 2025) listed the same price, effective at GA. | Launch | $0.075 / $0.3 | source |
| 02-24 | Claude 3.7 Sonnet Anthropic Same price in standard and extended-thinking modes; thinking tokens billed as output. | Launch | $3 / $15 | source |
| 02-18 | Gemini 1.0 Pro Google Changelog: 'Gemini 1.0 Pro is no longer supported.' The Feb 2025 pricing page gave Feb 15 as the shutdown date. | Retired | source | |
| 02-08 | DeepSeek-V3 DeepSeek Launch promotion ended; V3 list price took effect. | Price increase+173% | $0.27 / $1.1 | source |
| 02-05 | Gemini 2.0 Flash Google GA release. Text/image/video input $0.10; audio input $0.70. One price at all context lengths. | Launch | $0.1 / $0.4 | source |
| 01-31 | o3-mini OpenAI Price from archived pricing page (Feb 1, 2025 snapshot). | Launch | $1.1 / $4.4 | source |
| 01-30 | Mistral Small 3 Mistral Cheaper than Small 24.09 ($0.20/$0.60). Same price kept by Small 3.1 and 3.2. | Launch | $0.1 / $0.3 | source |
| 01-20 | DeepSeek-R1 DeepSeek deepseek-reasoner. Cache-hit input $0.14. | Launch | $0.55 / $2.19 | source |
| 01-13 | Codestral 25.01 Mistral Higher than the $0.20/$0.60 Codestral it replaced. | Launch | $0.3 / $0.9 | source |
2024
| 12-26 | DeepSeek-V3 DeepSeek Launched at V2 prices as a promotion until 2025-02-08; list price $0.27/$1.10 announced the same day. | Launch | $0.14 / $0.28 | source |
| 12-17 | o1 OpenAI o1 released in the API (changelog Dec 17, 2024). Price from archived pricing page (Jan 9, 2025 snapshot). | Launch | $15 / $60 | source |
| 12-12 | grok-2-1212 xAI xAI said it was lowering prices with this release. | Launch | $2 / $10 | source |
| 12-05 | Claude 3.5 Haiku Anthropic Cut 20% from $1 / $5 (date per Simon Willison's post quoting Anthropic, 2024-12-05). New price shown on the pricing page archived 2024-12-20. | Price cut−20% | $0.8 / $4 | source |
| 11-06 | Claude 1 / Claude Instant Anthropic claude-1.x and claude-instant-1.x retired (deprecated 2024-09-04). | Retired | source | |
| 11-04 | grok-beta xAI API public beta (announced 2024-11-04). Price from the earliest archived docs listing (2024-12-17). | Launch | $5 / $15 | source |
| 11-04 | Claude 3.5 Haiku Anthropic Released 2024-11-04 at 4x the Claude 3 Haiku price. Price as shown on the pricing page archived 2024-11-11. | Launch | $1 / $5 | source |
| 10-16 | Ministral 8B Mistral ministral-8b-2410. | Launch | $0.1 / $0.1 | source |
| 10-16 | Ministral 3B Mistral ministral-3b-2410. | Launch | $0.04 / $0.04 | source |
| 10-03 | Gemini 1.5 Flash-8B Google Prompts under 128K tokens; $0.075 / $0.30 above 128K. Paid billing started Oct 14, 2024. | Launch | $0.037 / $0.15 | source |
| 10-01 | Gemini 1.5 Pro Google Announced Sep 24, 2024 (64% input / 52% output cut for prompts under 128K); $2.50 / $10.00 above 128K. | Price cut−58% | $1.25 / $5 | source |
| 09-17 | Mistral Small Mistral 80% cut, with new Mistral Small 24.09. | Price cut−80% | $0.2 / $0.6 | source |
| 09-17 | Mistral Large 2 Mistral 33% cut. Same price kept by Mistral Large 24.11 (2024-11-18). | Price cut−33% | $2 / $6 | source |
| 09-17 | Codestral Mistral 80% cut. | Price cut−80% | $0.2 / $0.6 | source |
| 09-12 | o1-mini OpenAI Price from archived pricing page (Oct 1, 2024 snapshot). | Launch | $3 / $12 | source |
| 09-12 | o1-preview OpenAI First reasoning model; reasoning tokens billed as output. Price from archived pricing page (Oct 1, 2024 snapshot). | Launch | $15 / $60 | source |
| 09-05 | DeepSeek-V2.5 DeepSeek deepseek-chat and deepseek-coder merged into V2.5; price unchanged. Cache-hit input $0.014 since 2024-08-02. | Launch | $0.14 / $0.28 | source |
| 08-12 | Gemini 1.5 Flash Google Input cut 78%, output cut 71% for prompts under 128K tokens (announced Aug 8, 2024); $0.15 / $0.60 above 128K. | Price cut−75% | $0.075 / $0.3 | source |
| 08-06 | GPT-4o OpenAI New snapshot gpt-4o-2024-08-06: input -50%, output -33% vs gpt-4o-2024-05-13. | Price cut−42% | $2.5 / $10 | source |
| 07-24 | Mistral Large 2 Mistral mistral-large-2407. Price from archived pricing page (2024-07-31). | Launch | $3 / $9 | source |
| 07-18 | GPT-4o mini OpenAI 15 cents / 60 cents per 1M tokens, more than 60% cheaper than GPT-3.5 Turbo. | Launch | $0.15 / $0.6 | source |
| 06-20 | Claude 3.5 Sonnet Anthropic Same price as Claude 3 Sonnet. The upgraded 3.5 Sonnet (2024-10-22) kept this price. | Launch | $3 / $15 | source |
| 05-29 | Codestral Mistral codestral-2405. Price from archived pricing page (2024-06-06). | Launch | $1 / $3 | source |
| 05-14 | Gemini 1.5 Flash Google Prompts up to 128K tokens; $0.70 / $2.10 above 128K. Billing started May 30, 2024. | Launch | $0.35 / $1.05 | source |
| 05-14 | Gemini 1.5 Pro Google New price for prompts up to 128K tokens ($7.00 / $21.00 above 128K). Announced at Google I/O; billing started May 30, 2024. | Price cut−50% | $3.5 / $10.5 | source |
| 05-13 | GPT-4o OpenAI Half the price of GPT-4 Turbo. Price from the archived pricing page (June 1, 2024 snapshot). | Launch | $5 / $15 | source |
| 05-06 | DeepSeek-V2 DeepSeek Served as deepseek-chat (32K context on API). CNY list: ¥1/¥2. | Launch | $0.14 / $0.28 | source |
| 05-05 | Mistral Small Mistral Halved. Exact date unknown: $2/$6 on 2024-04-17 snapshot, $1/$3 on 2024-05-05 snapshot. | Price cut−50% | $1 / $3 | source |
| 05-05 | Mistral Large Mistral Halved. Exact date unknown: $8/$24 on 2024-04-17 snapshot, $4/$12 on 2024-05-05 snapshot. | Price cut−50% | $4 / $12 | source |
| 04-09 | Gemini 1.5 Pro Google Public preview in the Gemini API on Apr 9, 2024; paid tier shown as 'preview pricing', billing slated for May 2, 2024. Replaced before billing began. | Launch | $7 / $21 | source |
| 03-04 | Claude 3 Haiku Anthropic Price announced with the Claude 3 family; model became available on 2024-03-13. | Launch | $0.25 / $1.25 | source |
| 03-04 | Claude 3 Sonnet Anthropic Claude 3 family announced. 200K context. | Launch | $3 / $15 | source |
| 03-04 | Claude 3 Opus Anthropic Claude 3 family announced. 200K context. | Launch | $15 / $75 | source |
| 02-26 | Mistral Small Mistral mistral-small-2402. Price from archived docs (2024-03-07). | Launch | $2 / $6 | source |
| 02-26 | Mistral Large Mistral mistral-large-2402. Price from archived docs (2024-03-07); EUR €7.3/€22. | Launch | $8 / $24 | source |
| 02-15 | Gemini 1.0 Pro Google Listed price cut to $0.000125 per 1K characters in / $0.000375 out (changed between the Feb 4 and Feb 15, 2024 captures). By April 2024 Google listed it as $0.50 / $1.50 per 1M tokens. Paid billing started May 14, 2024. | Price cut−40% | $0.5 / $1.5 | source |
| 01-25 | GPT-3.5 Turbo OpenAI gpt-3.5-turbo-0125: input -50% to $0.0005, output -25% to $0.0015 per 1K tokens. | Price cut−40% | $0.5 / $1.5 | source |
| 01-18 | Gemini 1.0 Pro Google Released free on Dec 13, 2023; paid tier listed as 'coming soon' at $0.00025 per 1K characters in / $0.0005 per 1K characters out. Converted with Google's documented ~4 characters per token. Earliest archived capture that shows the price table (Dec 2023 captures do not render it). | Launch | $1 / $2 | source |
| 01-04 | GPT-3 davinci OpenAI davinci and text-davinci-001/002/003 shut down; replaced by davinci-002 and gpt-3.5-turbo-instruct. | Retired | source |
2023
| 12-12 | Claude Instant 1.2 Anthropic Cut from $1.63 / $5.51 (about -51% in, -56% out). Date = first archived copy of the new price sheet. | Price cut | $0.8 / $2.4 | source |
| 12-11 | Mistral Medium Mistral La Plateforme launch; priced in EUR only (€2.5/€7.5). USD list $2.7/$8.1 shown by 2024-03-07. | Launch | source | |
| 11-21 | Claude 2 Anthropic Cut from $11.02 / $32.68 the day Claude 2.1 launched (-27%). | Price cut−27% | $8 / $24 | source |
| 11-21 | Claude 2.1 Anthropic Launched with 200K context, at a lower price than Claude 2 had. | Launch | $8 / $24 | source |
| 11-06 | GPT-3.5 Turbo OpenAI New gpt-3.5-turbo-1106 (16K context) at $0.001 / $0.002 per 1K tokens. | Price cut−23% | $1 / $2 | source |
| 11-06 | GPT-4 Turbo OpenAI gpt-4-1106-preview, 128K context: $0.01 / $0.03 per 1K tokens, 3x cheaper input and 2x cheaper output than GPT-4. | Launch | $10 / $30 | source |
| 07-11 | Claude 2 Anthropic Same price as Claude 1. 100K context. | Launch | $11.02 / $32.68 | source |
| 06-13 | GPT-3.5 Turbo OpenAI Input cut 25% to $0.0015 per 1K tokens; output unchanged at $0.002. | Price cut−19% | $1.5 / $2 | source |
| 03-14 | Claude Instant 1 Anthropic Launched priced per character (about $0.43 in / $1.45 out per 1M characters). Anthropic's May 2023 price sheet states $1.63 / $5.51 per 1M tokens. | Launch | $1.63 / $5.51 | source |
| 03-14 | Claude 1 Anthropic Launched priced per character: $2.90 in / $8.60 out per 1M characters (Ars Technica). Anthropic's May 2023 price sheet states the same price per token: $11.02 / $32.68 per 1M tokens. | Launch | $11.02 / $32.68 | source |
| 03-14 | GPT-4 32K OpenAI gpt-4-32k: $0.06 / $0.12 per 1K tokens. | Launch | $60 / $120 | source |
| 03-14 | GPT-4 OpenAI gpt-4 8K context: $0.03 / $0.06 per 1K tokens. | Launch | $30 / $60 | source |
| 03-01 | GPT-3.5 Turbo OpenAI gpt-3.5-turbo (ChatGPT API) at $0.002 per 1K tokens, 10x cheaper than davinci. | Launch | $2 / $2 | source |
2022
Today's prices for every model, with a cost calculator: AI model API pricing. By provider: Claude API, OpenAI API, Gemini API, Grok API, Mistral API, DeepSeek API.
FAQ
Have AI API prices gone down?+
Yes, overall. The most expensive launch in this record is GPT-4.5 (OpenAI) at $75 input / $150 output per 1M tokens (Feb 27, 2025); the cheapest is Ministral 3B (Mistral) at $0.04 / $0.04 (Oct 16, 2024). Of the 28 list-price changes we found, 25 were cuts and 3 were increases.
What was the biggest AI API price cut?+
o3 (OpenAI) on Jun 10, 2025: from $10 / $40 to $2 / $8 per 1M tokens (input / output), −80% on a blended price.
How is the blended price calculated?+
Blended price = (3 × input price + output price) / 4, per 1M tokens: it assumes three input tokens for every output token, a common mix for chat and retrieval apps. The tables show input and output prices separately too.
Where do these prices come from?+
Each row links its source: the provider's pricing page (often an archived copy on the Wayback Machine), its announcement post or its docs. Prices are list prices in US dollars per 1M tokens, standard tier, base context length, without batch or cache discounts. Today's prices for every model are on the AI model pricing page; when its monthly re-check finds a change, it is added here with the check date.
Can I use this data?+
Yes. Download the CSV and cite "DevHunt AI API price history" with a link to this page.
Cite or embed this chart
Free to use with a link: “Source: DevHunt AI API price history, devhunt.org/ai-model-pricing/history”. The image updates when prices change.
[](https://devhunt.org/ai-model-pricing/history)