Kimi K3vsGPT-5.6 Sol
Across 4 shared benchmarks, GPT-5.6 Sol leads overall: Kimi K3 wins 0, GPT-5.6 Sol wins 4, with 0 ties and an average score difference of -8.60.
Kimi K3
Moonshot AI · 2026-07-16 · Reasoning model
GPT-5.6 Sol
OpenAI · 2026-06-26 · Reasoning model
Kimi K30 wins(0%)(100%)4 winsGPT-5.6 Sol
Benchmark scores
Grouped by capability, sorted by largest gap within each. 4 shared benchmarks.
AI Agent - Tool Usage
GPT-5.6 Sol 3/3| Benchmark | Kimi K3 | GPT-5.6 Sol | Diff |
|---|---|---|---|
| Agents' Last Exam | 28.304 / 4Max (With Tools) | 52.701 / 4极高强度思考（工具） | -24.40 |
| OSWorld 2.0 | 58.303 / 3Max (With Tools) | 62.602 / 3极高强度思考（工具） | -4.30 |
| TerminalBench 2.1 | 88.302 / 27Max (With Tools) | 88.801 / 27最高（无工具） | -0.50 |
Coding and Software Engineer
GPT-5.6 Sol 1/1| Benchmark | Kimi K3 | GPT-5.6 Sol | Diff |
|---|---|---|---|
| DeepSWE | 67.505 / 19Max (With Tools) | 72.701 / 19极高强度思考（工具） | -5.20 |
Specs
| Field | Kimi K3 | GPT-5.6 Sol |
|---|---|---|
| Publisher | Moonshot AI | OpenAI |
| Release date | 2026-07-16 | 2026-06-26 |
| Model type | Reasoning model | Reasoning model |
| Architecture | MoE | Dense |
| Parameters | 2.8T | Not available |
| Context length | 1M | 1.05M |
| Max output | 1M | 128K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | Kimi K3 | GPT-5.6 Sol |
|---|---|---|
| Text input | ¥20 / 1M tokens | $5 / 1M tokens |
| Text output | ¥100 / 1M tokens | $30 / 1M tokens |
| Cache read | ¥2 / 1M tokens | $0.5 / 1M tokens |
| Cache write | Not public | $6.25 / 1M tokens |
Summary
- GPT-5.6 Solleads in:AI Agent - Tool Usage (3/3), Coding and Software Engineer (1/1)
On average across the 4 shared benchmarks, GPT-5.6 Sol scores 8.60 higher.
Largest single-benchmark gap: Agents' Last Exam — Kimi K3 28.30 vs GPT-5.6 Sol 52.70 (-24.40).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.