Model comparison
Jamba Large vs Qwen3.8 Max
Qwen3.8 Max has enough public results to be ranked (#21); Jamba Large does not yet, so treat this comparison as directional.
Last verified . 0 shared benchmarks.
Alibaba (Qwen)Summary
- The widest gap is in reasoning, where Qwen3.8 Max leads 53.4 to 17.4.
- Qwen3.8 Max is cheaper at $2 / $6 per million input/output tokens, against $2 / $8 for Jamba Large.
- Qwen3.8 Max accepts more context: 1M tokens versus 256K.
- Jamba Large has downloadable open weights; the other is API-only.
Side by side
| Jamba Large | Qwen3.8 Max | |
|---|---|---|
| Provider | AI21 Labs | Alibaba (Qwen) |
| Noometry Index | 32.6 | 56.0 |
| Released | 2025-07-01 | 2026-08-02 |
| Weights | Open | Proprietary |
| Context window | 256K | 1M |
| Max output | 4K | 131K |
| Input $ / M tokens | $2 | $2 |
| Output $ / M tokens | $8 | $6 |
| Results tracked | 2 | 38 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Not comparable
Jamba Large: —, Qwen3.8 Max: 52.3 (#30)
| Benchmark | Jamba Large | Qwen3.8 Max |
|---|---|---|
| DeepSWE | — | 57.5% |
| LMArena WebDev | — | 1671 |
| FrontierSWE | — | 15.8% |
| SciCode | — | 52.9% |
| LMArena Coding | — | 1499 |
Agentic & Tool Use Not comparable
Jamba Large: —, Qwen3.8 Max: 42.6 (#17)
| Benchmark | Jamba Large | Qwen3.8 Max |
|---|---|---|
| τ²-bench Banking | — | 55.1% |
| GDP.pdf | — | 23.2% |
Reasoning Qwen3.8 Max leads
Jamba Large: 17.4, Qwen3.8 Max: 53.4 (#26)
| Benchmark | Jamba Large | Qwen3.8 Max |
|---|---|---|
| Kagi LLM Benchmark | 26.1% | — |
| NYT Connections (extended) | — | 88.3% |
| CritPt | — | 20% |
| Chess Puzzles | — | 40% |
| LMArena Hard Prompts | — | 1493 |
| Mystery Game Puzzles | — | 38% |
| DTBench | — | 92% |
| LMCA | — | 46.2% |
| Epoch Capabilities Index | — | 156.69 |
Math Not comparable
Jamba Large: —, Qwen3.8 Max: 73.1 (#15)
| Benchmark | Jamba Large | Qwen3.8 Max |
|---|---|---|
| FrontierMath (Tiers 1-3) | — | 74.7% |
| FrontierMath Tier 4 | — | 46.3% |
| OTIS Mock AIME 2024-2025 | — | 100% |
| ProofBench | — | 58% |
| LMArena Math | — | 1498 |
Knowledge Qwen3.8 Max leads
Jamba Large: 37.0, Qwen3.8 Max: 61.7 (#25)
| Benchmark | Jamba Large | Qwen3.8 Max |
|---|---|---|
| GPQA Diamond | — | 92.7% |
| SimpleQA Verified | — | 47.3% |
| Vectara Hallucination Rate | 9.7% | — |
| LMArena Expert | — | 1504 |
Multimodal Not comparable
Jamba Large: —, Qwen3.8 Max: 36.9 (#70)
| Benchmark | Jamba Large | Qwen3.8 Max |
|---|---|---|
| LMArena Vision | — | 1314 |
| Furniture Assembly | — | 20% |
Multilingual Not comparable
Jamba Large: —, Qwen3.8 Max: 56.6 (#18)
| Benchmark | Jamba Large | Qwen3.8 Max |
|---|---|---|
| LMArena Non-English | — | 1468 |
| LMArena Chinese | — | 1537 |
| LMArena French | — | 1502 |
| LMArena German | — | 1489 |
| LMArena Japanese | — | 1466 |
| LMArena Korean | — | 1467 |
| LMArena Russian | — | 1478 |
| LMArena Spanish | — | 1495 |
Instruction Following Not comparable
Jamba Large: —, Qwen3.8 Max: 77.7 (#19)
| Benchmark | Jamba Large | Qwen3.8 Max |
|---|---|---|
| LMArena Instruction Following | — | 1478 |
Long Context Not comparable
Jamba Large: —, Qwen3.8 Max: 45.5 (#30)
| Benchmark | Jamba Large | Qwen3.8 Max |
|---|---|---|
| LMArena Longer Query | — | 1486 |
Writing & Preference Not comparable
Jamba Large: —, Qwen3.8 Max: 67.0 (#31)
| Benchmark | Jamba Large | Qwen3.8 Max |
|---|---|---|
| LMArena Text | — | 1480 |
| LMArena Creative Writing | — | 1478 |
| LMArena Multi-Turn | — | 1487 |
Frequently asked questions
Is Jamba Large better than Qwen3.8 Max?
Qwen3.8 Max has enough public results to be ranked (#21); Jamba Large does not yet, so treat this comparison as directional.
Which is cheaper, Jamba Large or Qwen3.8 Max?
Qwen3.8 Max is cheaper. It lists at $2 per million input tokens and $6 per million output tokens; Jamba Large lists at $2 and $8.
Which has the bigger context window?
Qwen3.8 Max does, with 1M tokens against 256K.
How many benchmarks do Jamba Large and Qwen3.8 Max share?
0 benchmarks have published results for both models. Jamba Large has 2 scored results on Noometry and Qwen3.8 Max has 38.