Updated September 27, 2026. Rank says Qwen 3.6 Max (preview) is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.
Both of these models will change. Get the price, version and retirement notices for the pair, each with its source. Follow model changes
Qwen 3.6 Max (preview) has the higher public score estimate, 54.97 versus 40.06, but the 90% score intervals overlap. Treat that as a lead, not a settled winner. 2 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.
Code generation, repair, and software-engineering tasks
Qwen 3.6 Max (preview)
Qwen 3.6 Max (preview) leads on the public coding lane, 46.9 to 35.7, with Supported evidence for both models, although the 90% intervals overlap.
Prompts that approach the documented context limit
Qwen3.5-122B-A10B
Qwen3.5-122B-A10B has the larger documented context window.
Tool use, computer use, and multi-step task completion
Not enough matched evidence
Qwen 3.6 Max (preview) is not ranked on the public lane for agentic, so no winner is named for agentic.
1K fresh input + 500 output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
50K fresh input + 3K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Choose a job from the LLM Selector's task catalog to see the category row it rests on, under the same basis rules as the table below. A directional row stays directional; choosing a job never creates a winner.
The same task catalog as the LLM Selector. Each job names the evidence surface it rests on; nothing here adds a new score.
Like-for-like · BenchAlign v5.7
Qwen 3.6 Max (preview) leads the like-for-like coding row, although the 90% intervals overlap.
Coding scores combine specific tasks and setups. Match the editor, harness, and effort to your workflow.
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.
Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.
Terminal-Bench 2.0Agentic
Normalized gap 16.0SuperGPQAKnowledge
Normalized gap 6.8Each row shows the public-lane category score for both models: the BenchAlign v5.7 lane for agentic, coding, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.
| Category | Qwen3.5-122B-A10B | Qwen 3.6 Max (preview) | Basis | Reading |
|---|---|---|---|---|
| Coding | 35.7Supported · #73/135 | 46.9Supported · #44/135 | Like-for-likeBenchAlign v5.7 lane · 1 vs 5 public rows | Qwen 3.6 Max (preview) leads · intervals overlap |
| Agentic | 23.1Estimated · #84/105 | Not ranked | Not comparableBenchAlign v5.7 lane · 3 vs 2 public rows | Not comparable |
| Reasoning | 49.8Unranked · 3 rankable rows | Not ranked | Not comparableProvisional lane · 1 vs 0 weighted rows | Not comparable |
| Multimodal | 57.0#34/50 | Not ranked | Not comparableProvisional lane · 1 vs 0 weighted rows | Not comparable |
| Knowledge | 41.5Supported · #86/158 | Not ranked | Not comparableBenchAlign v5.7 lane · 3 vs 1 public rows | Not comparable |
| Multilingual | 36.8#10/12 | Not ranked | Not comparableProvisional lane · 1 vs 0 weighted rows | Not comparable |
| Instruction following | 91.6#11/124 | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Math | Not ranked | 41.3Unranked · 2 rankable rows | Not comparableProvisional lane · 0 vs 2 weighted rows | Not comparable |
Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign v5.7) differ from the provisional-lane categories. Unranked scores sit on the provisional lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.
Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.
1K fresh input + 500 output tokens
Qwen3.5-122B-A10B has no comparable published API token rate. Qwen 3.6 Max (preview) has no comparable published API token rate.
50K fresh input + 3K output tokens
Qwen3.5-122B-A10B has no comparable published API token rate. Qwen 3.6 Max (preview) has no comparable published API token rate.
200K cached + 20K fresh input + 10K output tokens
Qwen3.5-122B-A10B has no comparable published API token rate. Qwen 3.6 Max (preview) has no comparable published API token rate.
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
Maximum documented context; output-token limits may be lower.
Qwen3.5-122B-A10B
262K
Qwen 3.6 Max (preview)
256K
Qwen3.5-122B-A10B
Not sourced
Qwen 3.6 Max (preview)
Not sourced
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Qwen3.5-122B-A10B
No comparable hosted API rate
Qwen 3.6 Max (preview)
No comparable hosted API rate
Qwen3.5-122B-A10B
Not sourced
Qwen 3.6 Max (preview)
Not sourced
Qwen3.5-122B-A10B
Not sourced
Qwen 3.6 Max (preview)
Not sourced
Qwen3.5-122B-A10B
Not sourced
Qwen 3.6 Max (preview)
Not sourced
Qwen3.5-122B-A10B
Reasoning
Qwen 3.6 Max (preview)
Reasoning
Qwen3.5-122B-A10B
Open Weight
Qwen 3.6 Max (preview)
Proprietary
Qwen3.5-122B-A10B
Open Weight
Qwen 3.6 Max (preview)
Proprietary
Qwen3.5-122B-A10B
2026-03-04
Qwen 3.6 Max (preview)
2026-04-20
Qwen 3.6 Max (preview) has the higher public score estimate, 54.97 versus 40.06, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Qwen 3.6 Max (preview) leads the public coding lane, 46.9 to 35.7, with Supported evidence for both models, although the 90% intervals overlap.
Qwen 3.6 Max (preview) is not ranked on the public lane for agentic tasks, so no winner is named for agentic tasks.
Both models do not have comparable published API token rates, so this page does not name a universal price winner.
Qwen3.5-122B-A10B has the larger documented context window: 262K, compared with 256K.
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Terminal-Bench 2.0
Qwen 3.6 Max (preview) leads this result
BrowseComp
Not directly comparable
OSWorld-Verified
Not directly comparable
QwenClawBench
Not directly comparable
SWE-bench Verified
Not directly comparable
SWE-bench Pro
Not directly comparable
SciCode
Not directly comparable
NL2Repo
Not directly comparable
Terminal-Bench 2.0
Not directly comparable
SWE-bench (Vals)
Not directly comparable
LongBench v2
Not directly comparable
MMMU
Not directly comparable
MMVU
Not directly comparable
MathVision
Not directly comparable
CharXiv
Not directly comparable
V*
Not directly comparable
MMLU-ProX
Not directly comparable
IFEval
Not directly comparable
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.
Last updated September 27, 2026
