AI Model Pricing Cheat Sheet
Every major AI model's API price in one place — input and output cost per million tokens, context window, and what each is best for. Ranked cheapest first. Prices are verified from each provider's official pricing page. Tap a column to re-sort.
| Model | Maker | Input$/1M | Output$/1M | Blended$/1M * | Context | Best for |
|---|---|---|---|---|---|---|
| Grok 4 Fast | Groksrc ✓ | $0.2 | $0.5 | $0.28 | 128K tokens | Cheaper, faster tier for everyday queries and high-volume use. |
| DeepSeek-V3.2 | DeepSeeksrc ✓ | $0.28 | $0.42 | $0.32 | 128K tokens | Near-frontier chat and coding quality at the lowest API price of any major provider. |
| Mistral Medium 3 | Le Chat (Mistral)src ✓ | $0.4 | $2 | $0.8 | 128K tokens | Aggressively priced mid-tier that punches above its weight for everyday work. |
| Gemini 2.5 Flash | Geminisrc ✓ | $0.3 | $2.5 | $0.85 | 1M tokens | Very fast and cheap for everyday and high-volume workloads. |
| Sonar | Perplexitysrc ✓ | $1 | $1 | $1 | Varies by model | Search-grounded answers with citations at a very low price — built for live-web Q&A. |
| Claude Haiku 4.5 | Claudesrc ✓ | $1 | $5 | $2 | 200K tokens | Fast and cheap for high-volume tasks while staying genuinely capable. |
| Mistral Large | Le Chat (Mistral)src ✓ | $2 | $6 | $3 | 128K tokens | Europe's strongest frontier model — solid reasoning and code with EU data residency. |
| Claude Sonnet 5 | Claudesrc ✓ | $2 | $10 | $4 | 200K tokens | New (Jun 2026) agent-focused workhorse — near-flagship quality at a steep discount.Introductory pricing through Aug 31, 2026; then $3 in / $15 out. |
| Gemini 3 Pro | Geminisrc ✓ | $2 | $12 | $4.5 | 1M tokens | Frontier reasoning with a 1M-token context — reads entire codebases or document sets in one pass. |
| GPT-5.4 | ChatGPTsrc ✓ | $2.5 | $15 | $5.63 | 128K tokens | Strong general-purpose frontier model at half the flagship price. |
| Sonar Pro | Perplexitysrc ✓ | $3 | $15 | $6 | Varies by model | Deeper multi-step research answers with larger context and more citations. |
| Grok 4 | Groksrc ✓ | $3 | $15 | $6 | 128K tokens | Strong frontier reasoning with native access to real-time X (Twitter) data. |
| Claude Opus 4.8 | Claudesrc ✓ | $5 | $25 | $10 | 200K tokens | The frontier flagship (May 2026) — top-tier deep reasoning and coding for long, complex, multi-step work; no long-context surcharge. |
| GPT-5.5 | ChatGPTsrc ✓ | $5 | $30 | $11.25 | 128K tokens | Current flagship — hard reasoning, coding, and agentic tasks. (GPT-5.6 is in limited preview.) |
No models match that search.
* Blended assumes a 3:1 input:output token mix (typical for chat/RAG) so you can rank models with one number — your real ratio will differ.
These are headline list prices for each maker's flagship models. Theyexclude prompt caching (often up to ~90% off), batch-API discounts (~50%), volume/committed-use tiers, long-context surcharges, and image/audio token costs — so a real bill can differ. Always confirm on the provider's official pricing page (the link on each row) before budgeting. Prices are human-verified, never machine-written, consistent with how we rank & disclose.
Compare full tool features →Which models are actually shipping? →