GPT-5.4 vs Kimi K3 vs Claude Sonnet 5.5
Claude Sonnet 5.5 comes out ahead, 72 to 65 and 65 on our weighted score, and it is the cheaper option too.
OpenAIGPT-5.4
65/100- ECI156.9
- Price$2.50 / $15.00
- Context1.05M
Moonshot AIKimi K3
65/100- ECI157.6
- Price$3.00 / $15.00
- Context1.05M
- Our pick
AnthropicClaude Sonnet 5.5
72/100- ECI165.2
- Price$2.00 / $10.00
- Context1M
Claude Sonnet 5.5 is our pick
Claude Sonnet 5.5 is the better all-round choice, scoring 72/100 against Kimi K3 (65) and GPT-5.4 (65). It leads on capability and price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
- CapabilityClaude Sonnet 5.5Capabilities Index (ECI): Claude Sonnet 5.5 165.2 · Kimi K3 157.6 · GPT-5.4 156.9
- Lowest priceClaude Sonnet 5.5Claude Sonnet 5.5 $4.00 · GPT-5.4 $5.63 · Kimi K3 $6.00 per 1M tokens (3:1 blend)
- Longest contextGPT-5.4 and Kimi K3GPT-5.4 1,050,000 · Kimi K3 1,048,576 · Claude Sonnet 5.5 1,000,000 tokens
- Widest inputsSame inputsGPT-5.4: Text, Images, PDFs · Kimi K3: Text, Images, Video · Claude Sonnet 5.5: Text, Images, PDFs
- Self-hostingKimi K3Publishes downloadable weights
| Measure | Weight | GPT-5.4 | Kimi K3 | Claude Sonnet 5.5 |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 50% | 87 | 88 | 97 |
| Price | 25% | 14 | 13 | 21 |
| Inputs & features | 15% | 80 | 80 | 80 |
| Context window | 10% | 61 | 61 | 60 |
| Overall | 100% | 65/100 | 65/100 | 72/100 |
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | GPT-5.4OpenAI | Kimi K3Moonshot AI | Claude Sonnet 5.5Anthropic |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 156.9 | 157.6 | 165.2 (best) |
| ECI rank | #16 of 148 | #13 of 148 | #3 of 148 (best) |
| GPQA DiamondGraduate-level science questions | 93.3% | 93.1% | 95.6% (best) |
| FrontierMath Tiers 1–3Research-level mathematics | 78.6% | 72.2% | 88.8% (best) |
| OTIS Mock AIME 2024–2025Competition mathematics | 97.8% | 97.2% | 100% (best) |
| SWE-bench VerifiedFixing real GitHub issues | 76.9% | — | — |
| SimpleQA VerifiedShort factual questions | 45.1% | 50.6% (best) | 46.5% |
| Price per million tokens | |||
| Input | $2.50 | $3.00 | $2.00 (best) |
| Output | $15.00 | $15.00 | $10.00 (best) |
| Cached input | $0.25 | $0.30 | $0.20 (best) |
| Blended (3:1) | $5.63 | $6.00 | $4.00 (best) |
| Long-context rate | Over 272K: $5.00 / $22.50 | Same rate | Same rate |
| Price source | Official OpenAI API | Official Moonshot AI API | Official Anthropic API |
| Limits | |||
| Context window | 1,050,000 tokens (best) | 1,048,576 tokens | 1,000,000 tokens |
| Max output | 128,000 tokens | 131,072 tokens (best) | 128,000 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | Yes | Yes |
| PDFs | Yes | No | Yes |
| Audio | No | No | No |
| Video | No | Yes | No |
| Reasoning | Yeslow · medium · high · xhigh | Yeslow · high · max | Yeslow · medium · high · xhigh · max |
| Tool calling | Yes | Yes | Yes |
| Structured output | Yes | Yes | Yes |
| Availability | |||
| Weights | Proprietary | Open | Proprietary |
| API model ID | gpt-5.4 | kimi-k3 | claude-sonnet-5-5 |
| API providers | 36 | 68 (best) | 23 |
| Released | Mar 5, 2026 | Jul 16, 2026 | Sep 28, 2026 |
| Knowledge cutoff | Aug 31, 2025 | — | Jun 2026 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- GPT-5.4$55.00
- Kimi K3$60.00
- Claude Sonnet 5.5$40.00
Which should you choose?
Which is better: GPT-5.4, Kimi K3 or Claude Sonnet 5.5?
Claude Sonnet 5.5 is the better all-round choice, scoring 72/100 against Kimi K3 (65) and GPT-5.4 (65). It leads on capability and price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
Which is cheaper, GPT-5.4, Kimi K3 or Claude Sonnet 5.5?
Claude Sonnet 5.5 is cheaper at $2.00 input / $10.00 output per million tokens (official Anthropic API price). GPT-5.4 costs $2.50 input / $15.00 output per million tokens (official OpenAI API price); Kimi K3 costs $3.00 input / $15.00 output per million tokens (official Moonshot AI API price). At a typical mix of three input tokens to one output token, that is $4.00 per million tokens for Claude Sonnet 5.5 versus $5.63 for GPT-5.4 (1.4× as much) and $6.00 for Kimi K3 (1.5× as much).
Which scores higher on benchmarks?
Claude Sonnet 5.5 scores higher on the Capabilities Index (ECI): Claude Sonnet 5.5 165.2 (#3 of 148), Kimi K3 157.6 (#13 of 148) and GPT-5.4 156.9 (#16 of 148). Their confidence ranges do not overlap (161.8–169.4 vs 154.9–160.4), so the gap is a real one. On individual benchmarks: GPQA Diamond — Claude Sonnet 5.5 95.6%, GPT-5.4 93.3%, Kimi K3 93.1%; FrontierMath Tiers 1–3 — Claude Sonnet 5.5 88.8%, GPT-5.4 78.6%, Kimi K3 72.2%; OTIS Mock AIME 2024–2025 — Claude Sonnet 5.5 100%, GPT-5.4 97.8%, Kimi K3 97.2%; SimpleQA Verified — Kimi K3 50.6%, Claude Sonnet 5.5 46.5%, GPT-5.4 45.1%.
Which is better for coding?
There are no published SWE-bench Verified results for Kimi K3 and Claude Sonnet 5.5 yet, so there is no like-for-like coding score. On overall capability, Claude Sonnet 5.5 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?
GPT-5.4 and Kimi K3 have the largest context windows (1,050,000 and 1,048,576 tokens), against 1,000,000 for Claude Sonnet 5.5. Maximum output per response: GPT-5.4 up to 128,000, Kimi K3 up to 131,072, Claude Sonnet 5.5 up to 128,000 tokens.
Which can read images, PDFs, audio or video?
GPT-5.4 accepts text, images and PDFs; Kimi K3 accepts text, images and video; Claude Sonnet 5.5 accepts text, images and PDFs. They handle the same number of input types.
Are any of these open source?
Kimi K3 publishes its weights and can be self-hosted; GPT-5.4 and Claude Sonnet 5.5 is proprietary.
Which is newer?
Claude Sonnet 5.5 is the newest, released Sep 28, 2026. Kimi K3 came out Jul 16, 2026; GPT-5.4 came out Mar 5, 2026. Knowledge cutoff: GPT-5.4 Aug 31, 2025, Claude Sonnet 5.5 Jun 2026.
How do you decide the winner?
Each model gets a 0–100 score on 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 Oct 4, 2026.
