Updated October 19, 2026. Rank says Qwen3.6-27B is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.
Qwen3.6-27B has the higher public point estimate, 48.57 versus 41.15. Their conditional score ranges overlap. These ranges do not establish rank confidence. 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
Qwen3.6-27B
Qwen3.6-27B has the higher public coding point estimate, 34.8 to 16.8, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Tool use, computer use, and multi-step task completion
Qwen3.6-27B
Qwen3.6-27B has the higher public agentic point estimate, 29.1 to 21.5, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Prompts that approach the documented context limit
Qwen3.6-27B
Qwen3.6-27B has the larger documented context window.
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
The page does not recommend a cost winner because at least one model cannot fit the stated workload in one request. Claude Haiku 4.5 does not fit this workload in one request. Qwen3.6-27B has no comparable published API token rate.
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 v5.8
Qwen3.6-27B has the higher coding point estimate. Conditional score ranges do not establish rank confidence.
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.
1 category rests on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.
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.
SWE-bench VerifiedCoding
Normalized gap 3.9Each row shows the public-lane category score for both models: the BenchAlign v5.8 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 | Claude Haiku 4.5 | Qwen3.6-27B | Basis | Reading |
|---|---|---|---|---|
| Agentic | 21.5Supported · #109/130 | 29.1Supported · #97/130 | Like-for-likeBenchAlign v5.8 lane · 3 vs 6 public rows | Qwen3.6-27B leads · intervals overlap |
| Coding | 16.8Supported · #127/151 | 34.8Supported · #80/151 | Like-for-likeBenchAlign v5.8 lane · 4 vs 7 public rows | Qwen3.6-27B leads · intervals overlap |
| Knowledge | 33.2Estimated · #143/186 | 46.8Estimated · #92/186 | Directional onlyBenchAlign v5.8 lane · 2 vs 6 public rows | Directional only |
| Reasoning | Not ranked | 78.1Unranked · 2 rankable rows | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Multimodal | Not ranked | 67.3#39/55 | Not comparableProvisional lane · 0 vs 2 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Instruction following | Not ranked | 82.4#55/148 | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Math | Not ranked | 65.9#8/12 | Not comparableProvisional lane · 0 vs 3 weighted rows | Not comparable |
Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign v5.8) 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
Qwen3.6-27B has no comparable published API token rate.
50K fresh input + 3K output tokens
Qwen3.6-27B has no comparable published API token rate.
200K cached + 20K fresh input + 10K output tokens
Claude Haiku 4.5 does not fit this workload in one request. Qwen3.6-27B 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.
Claude Haiku 4.5
Qwen3.6-27B
262K
Claude Haiku 4.5
claude-haiku-4-5-20251001
Claude API pricingQwen3.6-27B
Not sourced
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Claude Haiku 4.5
$0.1 per 1M cached input tokens
Claude API pricingQwen3.6-27B
No comparable hosted API rate
Claude Haiku 4.5
Not sourced
Qwen3.6-27B
Not sourced
Claude Haiku 4.5
Not sourced
Qwen3.6-27B
Not sourced
Claude Haiku 4.5
Not sourced
Qwen3.6-27B
Not sourced
Claude Haiku 4.5
Non-Reasoning
Qwen3.6-27B
Reasoning
Claude Haiku 4.5
Proprietary
Qwen3.6-27B
Open Weight
Claude Haiku 4.5
Proprietary
Qwen3.6-27B
Open Weight
Claude Haiku 4.5
2025-10-15
Qwen3.6-27B
2026-04-21
Qwen3.6-27B has the higher public point estimate, 48.57 versus 41.15. Their conditional score ranges overlap. These ranges do not establish rank confidence. The page therefore keeps the decision tied to the specific documented workload.
Qwen3.6-27B has the higher public coding point estimate, 34.8 to 16.8, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Qwen3.6-27B has the higher public agentic tasks point estimate, 29.1 to 21.5, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Both models do not have comparable published API token rates, so this page does not name a universal price winner.
Qwen3.6-27B has the larger documented context window: 262K, compared with 200K.
Estimates at 50,000 req/day · 1000 tokens/req average.
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
JobBench
Not directly comparable
Terminal-Bench 2.1 (Vals)
Not directly comparable
MazeBench
Not directly comparable
Terminal-Bench 2.0
Not directly comparable
Claw-Eval
Not directly comparable
QwenClawBench
Not directly comparable
QwenWebBench
Not directly comparable
AndroidWorld
Not directly comparable
Gert Labs
Not directly comparable
SWE-bench Verified
Qwen3.6-27B leads this result
VulcanBench v3
Not directly comparable
LiveCodeBench (Vals)
Not directly comparable
SWE-bench (Vals)
Qwen3.6-27B leads this result
SWE Multilingual
Not directly comparable
SWE-bench Pro
Not directly comparable
Terminal-Bench 2.0
Not directly comparable
LiveCodeBench v6
Not directly comparable
NL2Repo
Not directly comparable
MMMU
Not directly comparable
MMMU-Pro
Not directly comparable
RealWorldQA
Not directly comparable
DynaMath
Not directly comparable
MStar
Not directly comparable
SimpleVQA
Not directly comparable
CharXiv
Not directly comparable
CC-OCR
Not directly comparable
CountBench
Not directly comparable
RefCOCO (avg)
Not directly comparable
ERQA
Not directly comparable
Video-MME (with subtitle)
Not directly comparable
VideoMMMU
Not directly comparable
MLVU (M-Avg)
Not directly comparable
V*
Not directly comparable
GPQA Diamond (Vals)
Not directly comparable
MMLU-Pro (Vals)
Not directly comparable
MMLU-Pro
Not directly comparable
MMLU-Redux
Not directly comparable
SuperGPQA
Not directly comparable
C-Eval
Not directly comparable
GPQA
Not directly comparable
HLE
Not directly comparable
HMMT Feb 2025
Not directly comparable
HMMT Nov 2025
Not directly comparable
HMMT Feb 2026
Not directly comparable
MMAnswerBench
Not directly comparable
AIME26
Not directly comparable
FrontierMath v2 (Tiers 1-3)
Not directly comparable
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 5,600+ readers.
Last updated October 19, 2026
