Research
LLM Tracker
Model pricing, context windows, and practical use cases.
This page compares large language models on the numbers that decide which one you ship: price per million tokens in and out, context window, benchmark scores, and what each is actually good for. I pick for fintech work — API integrations, compliance docs, support tooling — so the recommendations skew that way. Prices come from each provider's own pricing page, linked per model, and the whole table is published as JSON.
Updated Aug 17, 2026
Best for fintech ops
Claude Sonnet 4.6
Near-Opus coding at 60% cost. Best value for payments integrations, compliance docs, API work.
Best for deep reasoning
Claude Opus 4.6
Strongest reasoning. Use for complex analysis, architecture decisions, audit work.
Best for high volume
GPT-4o-mini
$0.15/MTok input. Use for classification, extraction, routing — bulk processing.
Best for long context
Gemini 2.5 Flash
1M context at $0.30/MTok. Cheapest way to process large documents.
Anthropic
OpenAI
Google
Meta
DeepSeek
xAI
Mistral
26 models · 12 frontier · 9 performance · 5 speed
| Model | Provider | Tier | Context | Input/MTok | Output/MTok | SWE-Bench Pro | Terminal-Bench | Released |
|---|---|---|---|---|---|---|---|---|
| Claude Fable 5 | Anthropic | FRONTIER | 1M | $10.00 | $50.00 | — | 83.8% | Jun 2026 |
| GPT-5.5 | OpenAI | FRONTIER | 1M | $5.00 | $30.00 | — | 83.1% | — |
| Claude Opus 4.8 | Anthropic | FRONTIER | 1M | $5.00 | $25.00 | — | 78.9% | — |
| Claude Opus 4.7 | Anthropic | FRONTIER | 1M | $5.00 | $25.00 | — | 68.9% | — |
| Gemini 3.1 Pro | FRONTIER | 1M | $2.00 | $12.00 | 46.1% | 65.8% | Feb 2026 | |
| GPT-5.4 | OpenAI | FRONTIER | 1M | $2.50 | $15.00 | 59.1% | — | Mar 2026 |
| Claude Opus 4.6 | Anthropic | FRONTIER | 1M | $5.00 | $25.00 | 51.9% | — | Feb 2026 |
| o3 | OpenAI | FRONTIER | 200K | $2.00 | $8.00 | — | — | Apr 2025 |
| DeepSeek V4 Pro | DeepSeek | FRONTIER | 1M | $0.43 | $0.87 | — | — | Apr 2026 |
| Grok 4.3 | xAI | FRONTIER | 1M | $1.25 | $2.50 | — | — | — |
| Mistral Medium 3.5 | Mistral | FRONTIER | 256K | $1.50 | $7.50 | — | — | Apr 2026 |
| Claude Opus 5 | Anthropic | FRONTIER | 1M | $5.00 | $25.00 | — | — | Jul 2026 |
| Claude Sonnet 5 | Anthropic | PERFORMANCE | 1M | $2.00 | $10.00 | — | 74.6% | Jun 2026 |
| Llama 4 Maverick | Meta | PERFORMANCE | 1M | $0.27 | $0.85 | 5.2% | — | Apr 2025 |
| Claude Sonnet 4.6 | Anthropic | PERFORMANCE | 1M | $3.00 | $15.00 | — | — | Feb 2026 |
| GPT-4o | OpenAI | PERFORMANCE | 128K | $2.50 | $10.00 | — | — | May 2024 |
| o3-mini | OpenAI | PERFORMANCE | 200K | $1.10 | $4.40 | — | — | Jan 2025 |
| Gemini 3.5 Flash | PERFORMANCE | 1M | $1.50 | $9.00 | — | — | — | |
| Gemini 2.5 Pro | PERFORMANCE | 1M | $1.25 | $10.00 | — | — | Mar 2025 | |
| DeepSeek V4 Flash | DeepSeek | PERFORMANCE | 1M | $0.14 | $0.28 | — | — | Apr 2026 |
| Mistral Large 3 | Mistral | PERFORMANCE | 256K | $0.50 | $1.50 | — | — | Dec 2025 |
| Claude Haiku 4.5 | Anthropic | SPEED | 200K | $1.00 | $5.00 | 39.5% | — | Oct 2025 |
| GPT-4o-mini | OpenAI | SPEED | 128K | $0.15 | $0.60 | — | — | Jul 2024 |
| Gemini 2.5 Flash | SPEED | 1M | $0.30 | $2.50 | — | — | Jun 2025 | |
| Llama 4 Scout | Meta | SPEED | 10M | $0.18 | $0.59 | — | — | Apr 2025 |
| Mistral Small 4 | Mistral | SPEED | 256K | $0.10 | $0.30 | — | — | Mar 2026 |
Click any column header to sort. Pricing from each provider's own docs. Scores come from Scale's SWE-Bench Pro public set and the tbench.ai Terminal-Bench 2.1 leaderboard, so they are comparable across rows and lag new releases. Where a maker publishes its own figure, the cell says so on hover.
Trolltunga, Norway
14-hour round trip. You earn that view.
