Agnes 2.5 Pro Theta vs Qwen3.5-9B
Full profiles: Agnes 2.5 Pro Theta (Sapiens AI) and Qwen3.5-9B (Alibaba): every benchmark, price and rank.
Updated · Data from 3 sources
As of 9 Nov 2026, Agnes 2.5 Pro Theta and Qwen3.5-9B are neck and neck on quality (50.8 vs 50.4 on the BenchLeader Index, inside the ±12.5 margin), and Qwen3.5-9B is 3.0× cheaper at $0.190 vs $0.563 per 1M tokens blended. Agnes 2.5 Pro Theta leads in every category we measure both in. Agnes 2.5 Pro Theta streams ~2× faster.
Key facts (as of 9 Nov 2026)
- Agnes 2.5 Pro Theta scores 50.8 on the BenchLeader Index and Qwen3.5-9B scores 50.4 (BenchLeader, 9 Nov 2026).
- Agnes 2.5 Pro Theta costs $0.563 per 1M tokens blended (list price, 9 Nov 2026).
- Qwen3.5-9B costs $0.190 per 1M tokens blended (list price, 9 Nov 2026).
- Agnes 2.5 Pro Theta streams ~2× faster than Qwen3.5-9B (Artificial Analysis, 7-day median, 9 Nov 2026).
- Of the 11 benchmarks both models have, Agnes 2.5 Pro Theta leads 11 (BenchLeader, 9 Nov 2026).
Agnes 2.5 Pro Theta benchmarks 
Qwen3.5-9B benchmarks Which to pick for…
The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.
| Use case | Pick | Why | Confidence |
|---|---|---|---|
| Reasoning | Agnes 2.5 Pro Theta | 44 vs 35 category score; HLE-Diamond (AA) 16.8% vs 4.8% | High |
| Long documents | Agnes 2.5 Pro Theta | 67 vs 60 category score; AA-LCR v1.1 84.7% vs 70.0% | High |
| Images and other media | Agnes 2.5 Pro Theta | 61 vs 51 category score; MMMU-Pro 78.4% vs 69.3% | High |
| Lowest cost | Either | $0.150 vs $0.155 per 1M blended; agentic coding session $0.007 vs $0.009 | – |
| Fastest first answer | Agnes 2.5 Pro Theta | 20.9 s (Agnes 2.5 Pro Theta) vs 40.2 s (Qwen3.5-9B (thinking)) to the first answer token | – |
| High volume on a budget | Neither flagship | A smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs | – |
Key differences
- Agnes 2.5 Pro Theta and Qwen3.5-9B (thinking) are neck and neck on quality (50.8 vs 50.4, within the ±12.5 margin).
- Agnes 2.5 Pro Theta is stronger in coding, knowledge, long context, multimodal, reasoning.
- Qwen3.5-9B (thinking) is stronger in agents & tools, instruction following.
- Qwen3.5-9B (thinking) is 3.0× cheaper ($0.190 vs $0.563 per 1M blended).
- Agnes 2.5 Pro Theta streams 2.5× faster (212 vs 86 tokens per second).
At a glance
| Spec | Agnes 2.5 Pro Theta | Qwen3.5-9B |
|---|---|---|
| Developer | Sapiens AI | Alibaba |
| Released | 7 Nov 2026 | 24 Feb 2026 |
| Context window (tokens) | 1,000,000 | 262,144 |
| Compared at | Agnes 2.5 Pro Theta | Qwen3.5-9B (thinking) |
| Open weights | No | Yes |
| Price in / out / cached input per 1M | $0.100 / $0.300 / – | $0.140 / $0.200 / $0.060 |
Category by category
Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.
| Category | Agnes 2.5 Pro Theta | Qwen3.5-9B (thinking) | Leader |
|---|---|---|---|
| Reasoning | 44.0 | 35.0 | Agnes 2.5 Pro Theta |
| Coding | 45.5 | – | not comparable |
| Agents & tools | – | 41.3 | not comparable |
| Knowledge | 38.6 | – | not comparable |
| Instruction following | – | 66.0 | not comparable |
| Multimodal | 60.9 | 51.4 | Agnes 2.5 Pro Theta |
| Long context | 67.1 | 59.5 | Agnes 2.5 Pro Theta |
On one chart
Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.
Not drawn, measured for fewer than two of these models: Coding, Agents & tools, Maths, Knowledge, Instruction following, Human preference.
Where sources disagree
Across the 11 benchmarks both have, Agnes 2.5 Pro Theta leads 11, Qwen3.5-9B leads 0.
| Overall index | Agnes 2.5 Pro Theta | Qwen3.5-9B (thinking) | Leader |
|---|---|---|---|
| BenchLeader Index | 50.8 | 50.4 | neck and neck |
Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).
Price, speed and latency
List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.
| Measure | Agnes 2.5 Pro Theta | Qwen3.5-9B (thinking) |
|---|---|---|
| Blended price per 1M tokens | $0.150 | $0.155 |
| Input / output per 1M | $0.100 / $0.300 | $0.140 / $0.200 |
| Cached input per 1M | – | $0.060 |
| Price per task: chat reply | $0.0001 | $0.0001 |
| Price per task: summarize a 30-page report | $0.0014 | $0.0018 |
| Price per task: code edit | $0.0011 | $0.0011 |
| Price per task: agentic coding session | $0.0072 | $0.0092 |
| Price per task: structured extraction | $0.0003 | $0.0003 |
| Agentic coding session, input mostly cached | – | $0.0056 |
| Cost per benchmark task (AA Index) | $0.575 | – |
| Output tokens per task (AA Index) | 278k | – |
| Of which reasoning | 229k | – |
| Time per task (AA Index) | 22 min | – |
| Output speed | 212 tok/s | 86 tok/s |
| Time to first answer | 20.9 s | 40.2 s |
| Time to a complete answer | 29.5 s | 102.8 s |
Each model is compared at its best-ranked reasoning-effort setting. Agnes 2.5 Pro Theta scores 50.8 and takes 21 s to its first answer token. Qwen3.5-9B (thinking) scores 50.4 and takes 40 s to its first answer token; Qwen3.5-9B (no reasoning) scores 49.2 in 1 s, Qwen3.5-9B scores 39.9 in 0 s. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.
Every benchmark
37 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.
| Metric | Publisher | Agnes 2.5 Pro Theta | Qwen3.5-9B (thinking) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 50.8 | 50.4 |
| Coding score | BenchLeader | 45.5 | – |
| Knowledge score | BenchLeader | 38.6 | – |
| Long context score | BenchLeader | 67.1 | 59.5 |
| Multimodal score | BenchLeader | 60.9 | 51.4 |
| Reasoning score | BenchLeader | 44.0 | 35.0 |
| Agents & tools score | BenchLeader | – | 41.3 |
| Instruction following score | BenchLeader | – | 66.0 |
| Blended price $/M | BenchLeader | $0.150 | $0.155 |
| Output speed | BenchLeader | 212 tok/s | 86 tok/s |
| Time to first answer | BenchLeader | 20.9 s | 40.2 s |
| Context window | BenchLeader | 1,000,000 tokens | 262,144 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | – | 79.0%† |
| GPQA Diamond (AA) | Artificial Analysis | – | 80.6% |
| Humanity's Last Exam (AA) | Artificial Analysis | 37.2% | 14.9% |
| HLE-Diamond (AA) | Artificial Analysis | 16.8% | 4.8% |
| CritPt | Artificial Analysis | 15.1% | 0.3% |
| CritPt v2 hard subset (AA) | Artificial Analysis | 26.0% | – |
| LMCA | Epoch AI Benchmarking Hub | – | 24.5%† |
| DTBench | Epoch AI Benchmarking Hub | – | 71.2%† |
| Coding | |||
| SciCode | SciCode | – | 29.5%† |
| SciCode (AA) | Artificial Analysis | 56.0% | 29.5% |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 10.1% | 0.5% |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | – | 29.2% |
| FrontierCode 2.0 (AA) | Artificial Analysis | 16.6% | – |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | – | 9.2%† |
| Terminal-Bench Hard | Artificial Analysis | – | 24.2% |
| τ²-Bench Telecom (AA) | Artificial Analysis | – | 86.8% |
| GDPval-AA v2.1, normalised score | Artificial Analysis | – | 0.0% |
| GDPval-AA v2.1 | Artificial Analysis | 1507 | 427 |
| τ³-Banking (AA) | Artificial Analysis | – | 7.0% |
| AutomationBench | Artificial Analysis | 48.8% | 1.1% |
| GDP.pdf (AA) | Artificial Analysis | 14.2% | 6.6% |
| AA-Briefcase v1.1 | Artificial Analysis | 1496 | 341 |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | 3.8% | – |
| Maths | |||
| OTIS Mock AIME | Epoch AI Benchmarking Hub | – | 42.8%† |
| AIME 2026 | MathArena | – | 92.5%† |
Show all 37 benchmarks
| Metric | Publisher | Agnes 2.5 Pro Theta | Qwen3.5-9B (thinking) |
|---|---|---|---|
| Maths | |||
| HMMT February 2026 | MathArena | – | 71.2%† |
| MathArena Apex | MathArena | – | 0.5%† |
| Knowledge | |||
| AA-Omniscience v1.1 | Artificial Analysis | -11.2 | – |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | 21.3% | – |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | 58.6% | – |
| Instruction following | |||
| IFBench | Artificial Analysis | – | 66.7% |
| Multimodal | |||
| MMMU-Pro | Artificial Analysis | 78.4% | 69.3% |
| Long context | |||
| AA-LCR v1.1 | Artificial Analysis | 84.7% | 70.0% |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | – | 139.3† |
| AA Intelligence Index v5.0 | Artificial Analysis | 24.0 | – |
| AA Intelligence Index v4.3.2 | Artificial Analysis | – | 11.2 |
| AA Progress Index v1 | Artificial Analysis | – | 239.6 |
Data as of 2026-11-09. Best configuration of each model; every score links to its source on the model pages.
What changed
- Agnes 2.5 Pro Theta: first results.
- Qwen3.5-9B (thinking): 1 new result (AA-Briefcase v1.1).
- Qwen3.5-9B (thinking): BenchLeader Index 51.0 → 50.5.
- Qwen3.5-9B (thinking): 2 new results (AA Progress Index v1, HLE-Diamond (AA)).
- Qwen3.5-9B (thinking): 2 new results (AutomationBench, GDP.pdf (AA)).
Related comparisons
- Agnes 2.5 Pro Theta vs GLM 4.7 Flash
- Agnes 2.5 Pro Theta vs Ling 3.1 Flash
- Qwen3.5-9B vs Nemotron 3.5 Super VL 120B A12B
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
Agnes 2.5 Pro Theta vs Qwen3.5-9B: questions
- Is Agnes 2.5 Pro Theta better than Qwen3.5-9B?
- Agnes 2.5 Pro Theta and Qwen3.5-9B (thinking) are neck and neck on quality (50.8 vs 50.4, within the ±12.5 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 9 Nov 2026, so check the category scores for your use.
- Agnes 2.5 Pro Theta vs Qwen3.5-9B: which is cheaper, and will that change?
- Qwen3.5-9B is cheaper today: $0.190 against $0.563 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Which is faster, Agnes 2.5 Pro Theta or Qwen3.5-9B?
- Agnes 2.5 Pro Theta streams ~2× faster (212 vs 86 output tokens per second, Artificial Analysis 7-day median).
- Which is better for long documents?
- Agnes 2.5 Pro Theta scores higher on long-context benchmarks (67 vs 60). Agnes 2.5 Pro Theta accepts more context: 1,000,000 against 262,144 tokens for Qwen3.5-9B.
- Which has lower latency?
- Agnes 2.5 Pro Theta answers first: 20.9 s against 40.2 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
More questions (2)
- Do independent benchmarks agree on Agnes 2.5 Pro Theta vs Qwen3.5-9B?
- Across the 11 benchmarks both have, Agnes 2.5 Pro Theta leads 11, Qwen3.5-9B leads 0.
- When was Agnes 2.5 Pro Theta released?
- Agnes 2.5 Pro Theta was released on 7 Nov 2026.