AI Model Comparison Tracker
Every frontier model in one sortable table — context window, API price per million tokens, coding benchmark standing, and release date. GPT, Claude, Gemini, Grok, and DeepSeek, verified against official provider pricing pages.
| Model | Provider | |||||
|---|---|---|---|---|---|---|
| GPT-5.6 Sol | OpenAI | Jul 9, 2026 | 1.05M | $5 | $30 | — |
| GPT-5.6 Terra | OpenAI | Jul 9, 2026 | 1.05M | $2.50 | $15 | — |
| GPT-5.6 Luna | OpenAI | Jul 9, 2026 | 1.05M | $1 | $6 | — |
| Claude Sonnet 5Intro pricing through Aug 31, 2026 — $3/$15 from Sep 1 | Anthropic | Jun 30, 2026 | 1M | $2 | $10 | 85.2% |
| Claude Fable 5Top coderAccess restored worldwide Jul 1, 2026 after US export controls were lifted | Anthropic | Jun 9, 2026 | 1M | $10 | $50 | 95.0% |
| Claude Opus 4.8 | Anthropic | May 28, 2026 | 1M | $5 | $25 | 88.6% |
| Gemini 3.5 Flash | May 19, 2026 | 1.05M | $1.50 | $9 | — | |
| DeepSeek V4 ProCache-miss input rate; cache hits $0.0036/1M. Open weights (MIT) | DeepSeek | Apr 24, 2026 | 1M | $0.435 | $0.87 | 80.6% |
| DeepSeek V4 FlashCheapestCache-miss input rate; cache hits $0.0028/1M | DeepSeek | Apr 24, 2026 | 1M | $0.14 | $0.28 | 79.0% |
| GPT-5.5Long-context rates (2× in, 1.5× out) apply above ~272K input tokens | OpenAI | Apr 23, 2026 | 1.05M | $5 | $30 | — |
| Grok 4.20Multi-Agent variant supports 2M context | xAI | Mar 10, 2026 | 1M | $1.25 | $2.50 | — |
| GPT-5.4 | OpenAI | Mar 5, 2026 | 1M | $2.50 | $15 | — |
| Gemini 3.1 Pro$4/$18 above 200K input tokens | Feb 19, 2026 | 1.05M | $2 | $12 | 80.6% | |
| Gemini 3 Flash | Dec 17, 2025 | 1.05M | $0.5 | $3 | 78.0% | |
| Claude Haiku 4.5 | Anthropic | Oct 1, 2025 | 200K | $1 | $5 | 73.3% |
Last verified July 14, 2026 against each provider's official pricing page. Prices are standard per-1M-token API rates — cached-input, batch, and long-context rates differ (see notes). *SWE-bench Verified, self-reported scores as tracked by llm-stats.com; OpenAI's newest models are not listed on that board, shown as “—”.
Want to know what these prices mean for your workload?
Open the cost calculator →Frequently asked questions
Which AI model is best right now?
As of July 14, 2026, Claude Fable 5 leads the SWE-bench Verified coding benchmark at 95.0%, ahead of Claude Opus 4.8 (88.6%) and Claude Sonnet 5 (85.2%). OpenAI's GPT-5.6 family (released July 9, 2026) is its newest flagship line but is not listed on the tracked leaderboard. "Best" depends on the job: Fable 5 tops raw capability, Sonnet 5 and GPT-5.6 Terra lead on price-to-performance, and DeepSeek V4 undercuts everyone on cost.
What is the cheapest frontier model API?
DeepSeek V4 Flash is the cheapest tracked model at $0.14 per 1M input tokens and $0.28 per 1M output tokens (cache-miss rate). Among Western providers, Gemini 3 Flash is the budget leader at $0.50/$3. The cheapest OpenAI option is GPT-5.6 Luna ($1/$6) and the cheapest Anthropic option is Claude Haiku 4.5 ($1/$5).
Which model has the largest context window?
Grok 4.20's Multi-Agent variant supports a 2M-token context window, the largest tracked. Most current flagships — GPT-5.6, Claude Fable 5, Claude Opus 4.8, Claude Sonnet 5, Gemini 3.1 Pro, and DeepSeek V4 — offer roughly 1M tokens. Claude Haiku 4.5 is the outlier at 200K.
How often is this comparison updated?
Every price, context window, benchmark score, and release date on this page is verified against the provider's official pricing page or announcement. It was last verified on July 14, 2026, and it is re-verified whenever a frontier model launches or a provider changes pricing.
Go deeper
- LLM API Cost Calculator — estimate your monthly spend
- Token Counter — count the tokens in a real prompt
- AI Timeline — every major AI event, month by month
- Build vs Buy Worksheet — should you build or buy this capability?
- LLM Rate Limits Reference — tiers, 429s & throughput levers
- What is an LLM?
- What is a context window?
- What is a token in an LLM?
Model launches change this table. Know first.
Original AI briefings for founders, PMs, and builders — free, every morning.