Claude Sonnet 4 vs Kimi K2 Instruct vs Gemini 2.5 Pro
Gemini 2.5 Pro comes out ahead, 64 to 53 and 52 on our weighted score, though Kimi K2 Instruct is 3.4× cheaper per token.
Price in rating: Included · Exclude price
Anthropic
Claude Sonnet 4
52/100- ECI141.9
- Price$3.00 / $15.00
- Context200K
Moonshot AI
Kimi K2 Instruct
53/100- ECI141.3
- Price$0.57 / $2.30
- Context131K
- Our pick
Google
Gemini 2.5 Pro
64/100- ECI144.9
- Price$1.25 / $10.00
- Context1.05M
Gemini 2.5 Pro is our pick
Gemini 2.5 Pro is the better all-round choice, scoring 64/100 against Kimi K2 Instruct (53) and Claude Sonnet 4 (52). It leads on capability, inputs & features and context window. Kimi K2 Instruct wins on price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
- CapabilityGemini 2.5 ProCapabilities Index (ECI): Gemini 2.5 Pro 144.9 · Claude Sonnet 4 141.9 · Kimi K2 Instruct 141.3
- Lowest priceKimi K2 InstructKimi K2 Instruct $1.00 · Gemini 2.5 Pro $3.44 · Claude Sonnet 4 $6.00 per 1M tokens (3:1 blend)
- Longest contextGemini 2.5 ProGemini 2.5 Pro 1,048,576 · Claude Sonnet 4 200,000 · Kimi K2 Instruct 131,072 tokens
- Widest inputsGemini 2.5 ProClaude Sonnet 4: Text, Images, PDFs · Kimi K2 Instruct: Text · Gemini 2.5 Pro: Text, Images, PDFs, Audio, Video
- Self-hostingKimi K2 InstructPublishes downloadable weights (Modified MIT)
| Measure | Weight | Claude Sonnet 4 | Kimi K2 Instruct | Gemini 2.5 Pro |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 50% | 70 | 69 | 73 |
| Price | 25% | 12 | 48 | 23 |
| Inputs & features | 15% | 70 | 25 | 100 |
| Context window | 10% | 39 | 32 | 65 |
| Overall | 100% | 52/100 | 53/100 | 64/100 |
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | |||
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 141.9 | 141.3 | 144.9 (best) |
| ECI rank | #104 of 158 | #108 of 158 | #92 of 158 (best) |
| GPQA DiamondGraduate-level science questions | 79.2% | — | 85.3% (best) |
| FrontierMath Tiers 1–3Research-level mathematics | — | — | 24.6% |
| OTIS Mock AIME 2024–2025Competition mathematics | 71.1% | — | 84.7% (best) |
| SWE-bench VerifiedFixing real GitHub issues | — | — | 57.6% |
| Price per million tokens | |||
| Input | $3.00 | $0.57 (best) | $1.25 |
| Output | $15.00 | $2.30 (best) | $10.00 |
| Cached input | — | — | $0.125 |
| Blended (3:1) | $6.00 | $1.00 (best) | $3.44 |
| Long-context rate | Same rate | Same rate | Over 200K: $2.50 / $15.00 |
| Price source | Median of 10 providers | Median of 11 providers | Official Google API |
| Limits | |||
| Context window | 200,000 tokens | 131,072 tokens | 1,048,576 tokens (best) |
| Max output | 64,000 tokens | 32,768 tokens | 65,536 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | No | Yes |
| PDFs | Yes | No | Yes |
| Audio | No | No | Yes |
| Video | No | No | Yes |
| Reasoning | Yes | No | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | Yes |
| Availability | |||
| Weights | Proprietary | OpenModified MIT | Proprietary |
| API model ID | — | — | gemini-2.5-pro |
| API providers | 11 | 11 | 31 (best) |
| Released | May 22, 2025 | Jul 11, 2025 | Jun 17, 2025 |
| Knowledge cutoff | Mar 31, 2025 | — | Jan 2025 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
Claude Sonnet 4$60.00
Kimi K2 Instruct$10.30
Gemini 2.5 Pro$32.50
Which should you choose?
Which is better: Claude Sonnet 4, Kimi K2 Instruct or Gemini 2.5 Pro?
Gemini 2.5 Pro is the better all-round choice, scoring 64/100 against Kimi K2 Instruct (53) and Claude Sonnet 4 (52). It leads on capability, inputs & features and context window. Kimi K2 Instruct wins on price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
Which is cheaper, Claude Sonnet 4, Kimi K2 Instruct or Gemini 2.5 Pro?
Kimi K2 Instruct is cheaper at $0.57 input / $2.30 output per million tokens (median across 11 API providers). Gemini 2.5 Pro costs $1.25 input / $10.00 output per million tokens (official Google API price); Claude Sonnet 4 costs $3.00 input / $15.00 output per million tokens (median across 10 API providers). At a typical mix of three input tokens to one output token, that is $1.00 per million tokens for Kimi K2 Instruct versus $3.44 for Gemini 2.5 Pro (3.4× as much) and $6.00 for Claude Sonnet 4 (6× as much).
Which scores higher on benchmarks?
Gemini 2.5 Pro scores higher on the Capabilities Index (ECI): Gemini 2.5 Pro 144.9 (#92 of 158), Claude Sonnet 4 141.9 (#104 of 158) and Kimi K2 Instruct 141.3 (#108 of 158). Their confidence ranges do not overlap (143.5–146.3 vs 140.3–143.4), so the gap is a real one.
Which is better for coding?
There are no published SWE-bench Verified results for Claude Sonnet 4 and Kimi K2 Instruct yet, so there is no like-for-like coding score. On overall capability, Gemini 2.5 Pro leads, which tends to carry over to coding, but test on your own codebase. All three support tool calling for agent workflows.
Which has the bigger context window?
Gemini 2.5 Pro has the largest context window at 1,048,576 tokens, against 200,000 for Claude Sonnet 4 and 131,072 for Kimi K2 Instruct. Maximum output per response: Claude Sonnet 4 up to 64,000, Kimi K2 Instruct up to 32,768, Gemini 2.5 Pro up to 65,536 tokens.
Which can read images, PDFs, audio or video?
Claude Sonnet 4 accepts text, images and PDFs; Kimi K2 Instruct accepts text; Gemini 2.5 Pro accepts text, images, PDFs, audio and video. Gemini 2.5 Pro handles the widest range of inputs.
Are any of these open source?
Kimi K2 Instruct publishes its weights (Modified MIT) and can be self-hosted; Claude Sonnet 4 and Gemini 2.5 Pro is proprietary.
Which is newer?
Kimi K2 Instruct is the newest, released Jul 11, 2025. Gemini 2.5 Pro came out Jun 17, 2025; Claude Sonnet 4 came out May 22, 2025. Knowledge cutoff: Claude Sonnet 4 Mar 31, 2025, Gemini 2.5 Pro Jan 2025.
How do you decide the winner?
This rating uses capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Nov 18, 2026.