Qwen3 Max vs Kimi K2 Instruct vs Claude Haiku 4.5
Claude Haiku 4.5 comes out ahead, 60 to 53 and 51 on our weighted score, though Kimi K2 Instruct is 2× cheaper per token.
Price in rating: Included · Exclude price
Alibaba (Qwen)
Qwen3 Max
51/100- ECI142.4
- Price$1.20 / $6.00
- Context262K
Moonshot AI
Kimi K2 Instruct
53/100- ECI141.3
- Price$0.57 / $2.30
- Context131K
- Our pick
Anthropic
Claude Haiku 4.5
60/100- ECI142.8
- Price$1.00 / $5.00
- Context200K
Claude Haiku 4.5 is our pick
Claude Haiku 4.5 is the better all-round choice, scoring 60/100 against Kimi K2 Instruct (53) and Qwen3 Max (51). It leads on inputs & features. Qwen3 Max wins on context window. Kimi K2 Instruct wins on price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
- CapabilityClaude Haiku 4.5Capabilities Index (ECI): Claude Haiku 4.5 142.8 · Qwen3 Max 142.4 · Kimi K2 Instruct 141.3
- Lowest priceKimi K2 InstructKimi K2 Instruct $1.00 · Claude Haiku 4.5 $2.00 · Qwen3 Max $2.40 per 1M tokens (3:1 blend)
- Longest contextQwen3 MaxQwen3 Max 262,144 · Claude Haiku 4.5 200,000 · Kimi K2 Instruct 131,072 tokens
- Widest inputsClaude Haiku 4.5Qwen3 Max: Text · Kimi K2 Instruct: Text · Claude Haiku 4.5: Text, Images, PDFs
- Self-hostingKimi K2 InstructPublishes downloadable weights (Modified MIT)
| Measure | Weight | Qwen3 Max | Kimi K2 Instruct | Claude Haiku 4.5 |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 50% | 70 | 69 | 70 |
| Price | 25% | 30 | 48 | 34 |
| Inputs & features | 15% | 25 | 25 | 80 |
| Context window | 10% | 43 | 32 | 39 |
| Overall | 100% | 51/100 | 53/100 | 60/100 |
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | |||
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 142.4 | 141.3 | 142.8 (best) |
| ECI rank | #103 of 158 | #108 of 158 | #102 of 158 (best) |
| GPQA DiamondGraduate-level science questions | 72.6% (best) | — | 71.2% |
| FrontierMath Tiers 1–3Research-level mathematics | 19.0% | — | — |
| OTIS Mock AIME 2024–2025Competition mathematics | 73.3% (best) | — | 66.7% |
| SimpleQA VerifiedShort factual questions | 48.8% (best) | — | 13.2% |
| Price per million tokens | |||
| Input | $1.20 | $0.57 (best) | $1.00 |
| Output | $6.00 | $2.30 (best) | $5.00 |
| Cached input | — | — | $0.10 |
| Blended (3:1) | $2.40 | $1.00 (best) | $2.00 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Official Alibaba API | Median of 11 providers | Official Anthropic API |
| Limits | |||
| Context window | 262,144 tokens (best) | 131,072 tokens | 200,000 tokens |
| Max output | 65,536 tokens (best) | 32,768 tokens | 64,000 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | No | Yes |
| PDFs | No | No | Yes |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | No | No | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | Yes |
| Availability | |||
| Weights | Proprietary | OpenModified MIT | Proprietary |
| API model ID | qwen3-max | — | claude-haiku-4-5 |
| API providers | 13 | 11 | 57 (best) |
| Released | Sep 23, 2025 | Jul 11, 2025 | Oct 15, 2025 |
| Knowledge cutoff | Apr 2025 | — | Feb 28, 2025 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
Qwen3 Max$24.00
Kimi K2 Instruct$10.30
Claude Haiku 4.5$20.00
Which should you choose?
Which is better: Qwen3 Max, Kimi K2 Instruct or Claude Haiku 4.5?
Claude Haiku 4.5 is the better all-round choice, scoring 60/100 against Kimi K2 Instruct (53) and Qwen3 Max (51). It leads on inputs & features. Qwen3 Max wins on context window. Kimi K2 Instruct wins on price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
Which is cheaper, Qwen3 Max, Kimi K2 Instruct or Claude Haiku 4.5?
Kimi K2 Instruct is cheaper at $0.57 input / $2.30 output per million tokens (median across 11 API providers). Claude Haiku 4.5 costs $1.00 input / $5.00 output per million tokens (official Anthropic API price); Qwen3 Max costs $1.20 input / $6.00 output per million tokens (official Alibaba API price). At a typical mix of three input tokens to one output token, that is $1.00 per million tokens for Kimi K2 Instruct versus $2.00 for Claude Haiku 4.5 (2× as much) and $2.40 for Qwen3 Max (2.4× as much).
Which scores higher on benchmarks?
Claude Haiku 4.5 scores higher on the Capabilities Index (ECI): Claude Haiku 4.5 142.8 (#102 of 158), Qwen3 Max 142.4 (#103 of 158) and Kimi K2 Instruct 141.3 (#108 of 158). The confidence ranges of the top two overlap (141.2–144.3 vs 140.3–144.5), so treat the gap as small.
Which is better for coding?
There are no published SWE-bench Verified results for Qwen3 Max, Kimi K2 Instruct and Claude Haiku 4.5 yet, so there is no like-for-like coding score. On overall capability, Claude Haiku 4.5 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?
Qwen3 Max has the largest context window at 262,144 tokens, against 200,000 for Claude Haiku 4.5 and 131,072 for Kimi K2 Instruct. Maximum output per response: Qwen3 Max up to 65,536, Kimi K2 Instruct up to 32,768, Claude Haiku 4.5 up to 64,000 tokens.
Which can read images, PDFs, audio or video?
Qwen3 Max accepts text; Kimi K2 Instruct accepts text; Claude Haiku 4.5 accepts text, images and PDFs. Claude Haiku 4.5 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; Qwen3 Max and Claude Haiku 4.5 is proprietary.
Which is newer?
Claude Haiku 4.5 is the newest, released Oct 15, 2025. Qwen3 Max came out Sep 23, 2025; Kimi K2 Instruct came out Jul 11, 2025. Knowledge cutoff: Qwen3 Max Apr 2025, Claude Haiku 4.5 Feb 28, 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.