Qwen3 235B-A22B Thinking 2507 vs Grok 4.3 vs Kimi K2 Instruct
Grok 4.3 comes out ahead, 67 to 57 and 53 on our weighted score, though Qwen3 235B-A22B Thinking 2507 is 39% cheaper per token.
Price in rating: Included · Exclude price
Alibaba (Qwen)
Qwen3 235B-A22B Thinking 2507
57/100- ECI143.3
- Price$0.30 / $2.90
- Context262K
- Our pick
xAI
Grok 4.3
67/100- ECI149.9
- Price$1.25 / $2.50
- Context1M
Moonshot AI
Kimi K2 Instruct
53/100- ECI141.3
- Price$0.57 / $2.30
- Context131K
Grok 4.3 is our pick
Grok 4.3 is the better all-round choice, scoring 67/100 against Qwen3 235B-A22B Thinking 2507 (57) and Kimi K2 Instruct (53). It leads on capability, inputs & features and context window. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
- CapabilityGrok 4.3Capabilities Index (ECI): Grok 4.3 149.9 · Qwen3 235B-A22B Thinking 2507 143.3 · Kimi K2 Instruct 141.3
- Lowest priceQwen3 235B-A22B Thinking 2507Qwen3 235B-A22B Thinking 2507 $0.95 · Kimi K2 Instruct $1.00 · Grok 4.3 $1.56 per 1M tokens (3:1 blend)
- Longest contextGrok 4.3Grok 4.3 1,000,000 · Qwen3 235B-A22B Thinking 2507 262,144 · Kimi K2 Instruct 131,072 tokens
- Widest inputsGrok 4.3Qwen3 235B-A22B Thinking 2507: Text · Grok 4.3: Text, Images, PDFs · Kimi K2 Instruct: Text
- Self-hostingQwen3 235B-A22B Thinking 2507 and Kimi K2 InstructPublishes downloadable weights (Apache 2.0 and Modified MIT)
| Measure | Weight | Qwen3 235B-A22B Thinking 2507 | Grok 4.3 | Kimi K2 Instruct |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 50% | 71 | 77 | 69 |
| Price | 25% | 49 | 39 | 48 |
| Inputs & features | 15% | 35 | 80 | 25 |
| Context window | 10% | 43 | 64 | 32 |
| Overall | 100% | 57/100 | 67/100 | 53/100 |
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | |||
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 143.3 | 149.9 (best) | 141.3 |
| ECI rank | #100 of 158 | #69 of 158 (best) | #108 of 158 |
| GPQA DiamondGraduate-level science questions | 80.1% | 88.8% (best) | — |
| FrontierMath Tiers 1–3Research-level mathematics | — | 42.8% | — |
| OTIS Mock AIME 2024–2025Competition mathematics | 86.7% | 93.3% (best) | — |
| SimpleQA VerifiedShort factual questions | 40.4% (best) | 33.2% | — |
| Price per million tokens | |||
| Input | $0.30 (best) | $1.25 | $0.57 |
| Output | $2.90 | $2.50 | $2.30 (best) |
| Cached input | — | $0.20 | — |
| Blended (3:1) | $0.95 (best) | $1.56 | $1.00 |
| Long-context rate | Same rate | Over 200K: $2.50 / $5.00 | Same rate |
| Price source | Median of 11 providers | Official xAI API | Median of 11 providers |
| Limits | |||
| Context window | 262,144 tokens | 1,000,000 tokens (best) | 131,072 tokens |
| Max output | 81,920 tokens (best) | 30,000 tokens | 32,768 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | Yes | No |
| PDFs | No | Yes | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | Yes | Yeslow · medium · high · xhigh | No |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | Yes | No |
| Availability | |||
| Weights | OpenApache 2.0 | Proprietary | OpenModified MIT |
| API model ID | — | grok-4.3 | — |
| API providers | 12 | 30 (best) | 11 |
| Released | Jul 25, 2025 | Apr 17, 2026 | Jul 11, 2025 |
| Knowledge cutoff | — | — | — |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
Qwen3 235B-A22B Thinking 2507$8.80
Grok 4.3$17.50
Kimi K2 Instruct$10.30
Which should you choose?
Which is better: Qwen3 235B-A22B Thinking 2507, Grok 4.3 or Kimi K2 Instruct?
Grok 4.3 is the better all-round choice, scoring 67/100 against Qwen3 235B-A22B Thinking 2507 (57) and Kimi K2 Instruct (53). It leads on capability, inputs & features and context window. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
Which is cheaper, Qwen3 235B-A22B Thinking 2507, Grok 4.3 or Kimi K2 Instruct?
Qwen3 235B-A22B Thinking 2507 is cheaper at $0.30 input / $2.90 output per million tokens (median across 11 API providers). Kimi K2 Instruct costs $0.57 input / $2.30 output per million tokens (median across 11 API providers); Grok 4.3 costs $1.25 input / $2.50 output per million tokens (official xAI API price). At a typical mix of three input tokens to one output token, that is $0.95 per million tokens for Qwen3 235B-A22B Thinking 2507 versus $1.00 for Kimi K2 Instruct (1.1× as much) and $1.56 for Grok 4.3 (1.6× as much).
Which scores higher on benchmarks?
Grok 4.3 scores higher on the Capabilities Index (ECI): Grok 4.3 149.9 (#69 of 158), Qwen3 235B-A22B Thinking 2507 143.3 (#100 of 158) and Kimi K2 Instruct 141.3 (#108 of 158). Their confidence ranges do not overlap (147.9–151.9 vs 141.4–145.1), so the gap is a real one.
Which is better for coding?
There are no published SWE-bench Verified results for Qwen3 235B-A22B Thinking 2507, Grok 4.3 and Kimi K2 Instruct yet, so there is no like-for-like coding score. On overall capability, Grok 4.3 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?
Grok 4.3 has the largest context window at 1,000,000 tokens, against 262,144 for Qwen3 235B-A22B Thinking 2507 and 131,072 for Kimi K2 Instruct. Maximum output per response: Qwen3 235B-A22B Thinking 2507 up to 81,920, Grok 4.3 up to 30,000, Kimi K2 Instruct up to 32,768 tokens.
Which can read images, PDFs, audio or video?
Qwen3 235B-A22B Thinking 2507 accepts text; Grok 4.3 accepts text, images and PDFs; Kimi K2 Instruct accepts text. Grok 4.3 handles the widest range of inputs.
Are any of these open source?
Qwen3 235B-A22B Thinking 2507 and Kimi K2 Instruct publishes its weights (Apache 2.0 and Modified MIT) and can be self-hosted; Grok 4.3 is proprietary.
Which is newer?
Grok 4.3 is the newest, released Apr 17, 2026. Qwen3 235B-A22B Thinking 2507 came out Jul 25, 2025; Kimi K2 Instruct came out Jul 11, 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.