Claude Opus 5 vs Kimi K3
Claude Opus 5 against Kimi K3 is the closed-versus-open frontier question of August 2026. Opus 5 leads the Intelligence Index 63 to 60; K3 is the strongest open-weights model ever measured, undercuts Opus on price at $3/$15 versus $5/$25 per million tokens, and can be self-hosted. Sovereignty needs decide this one.
Side by side
| Claude Opus 5 | Kimi K3 | |
|---|---|---|
| Maker | Anthropic | Moonshot AI |
| Released | 2026-07-24 | 2026-07-16 |
| Context window | 1M tokens | 1M tokens |
| Modalities in | text, image | text, image |
| Input $/MTok | $5 | $3 |
| Output $/MTok | $25 | $15 |
| Open weights | No | Yes (Kimi K3 License (commercial terms apply)) |
Benchmark deltas
| Benchmark | Claude Opus 5 | Kimi K3 | Δ |
|---|---|---|---|
| AA Intelligence Index | 63 | 60 | +3 |
Δ is Claude Opus 5 minus Kimi K3, in points. Shaded cell wins the row. Dashed values are vendor-published, not yet reproduced — methodology.
Cost calculator
- Claude Opus 5
- $500/mo
- Kimi K3
- $300/mo
- Difference
- $200/mo
List prices as of 2026-08-11; excludes caching, batch, and long-context tiers — see model pages for tier notes.
Verdict
Three Intelligence Index points now separate the best closed model from the best open one — the narrowest that gap has ever measured. Opus 5 keeps the capability crown and the mature API ecosystem. K3's 2.8-trillion-parameter MoE delivers 95% of the measured intelligence at 60% of the price, with downloadable weights for data-residency, fine-tuning, and cost-ceiling requirements no closed API can meet — though its commercial license requires review, and 104B active parameters make self-hosting a serious infrastructure commitment.
Verdicts are analysis, not measurement, and are labeled as such per our editorial policy.
FAQ
- Is Claude Opus 5 cheaper than Kimi K3?
- No. Kimi K3 is cheaper: $3 input / $15 output per million tokens against $5 / $25 for Claude Opus 5 (prices as of August 11, 2026).
- Which has the longer context window — Claude Opus 5 or Kimi K3?
- Kimi K3 does — 1M tokens against 1M for Claude Opus 5.
- Which is the more capable model, Claude Opus 5 or Kimi K3?
- Claude Opus 5 scores higher on the Artificial Analysis Intelligence Index: 63 (measured August 11, 2026) against 60 for Kimi K3 (measured August 11, 2026).
Related news
Comparison published August 11, 2026; underlying data last changed August 11, 2026.