VENICE ATLAS · MODEL DOSSIER · QWEN 3.8 27B

Model overview

Qwen 3.8 27B

Name: Qwen 3.8 27B
Brand: venice.ai
Price: 0.4500 USD
Availability: InStock

Qwen 3.8 27B is a native vision-language dense model with 27B parameters. It improves coding, professional work, research, and long-horizon agentic tasks, with flexible thinking control and image and video understanding. It supports a native 262K-token context window.

qwen-3-8-27b · venice.ai

Text · Venice catalog$0.45 per 1M tokensPrice on Venice AI

ModalityTextOn Venice AI

Primary price$0.45 per 1M tokensper 1M tokens

Context window262KMax tokens

Benchmarks26Benchmark results

Benchmark results

See the tests behind each score. They use the benchmark’s original scoring scale.

Agents#3729.53 rating Biology#2830.53 rating Chemistry#3130.53 rating Vision#1933.08 rating General intelligence#3644.43 rating Instruction Following#925.32 rating Multimodal#736.53 rating Physics#3130.53 rating Productivity#125.06 rating Reasoning#3844.18 rating Spatial Reasoning#1724.56 rating Structured Output#1024.38 rating

Agents7 results

Agents' Last Examagents-last-exam · agents

42.9%

ReportedOct 2026

AndroidWorldandroidworld · agents

81.9%

ReportedOct 2026

ClawEval-MMclaw-eval-mm · agents

56.9%

ReportedOct 2026

CoWorkBenchcoworkbench · agents

70.7%

ReportedOct 2026

Job Benchjob-bench · agents

33.4%

ReportedOct 2026

RecreationBenchrecreationbench · agents

47.1%

ReportedOct 2026

WebArena-Verifiedwebarena-verified · agents

64.8%

ReportedOct 2026

Coding5 results

DeepSWE 1.1deepswe-1.1 · code

42.2%

ReportedOct 2026

NL2Reponl2repo · code

42.3%

ReportedOct 2026

QwenSWEBenchqwen-swe-bench · code

79.0%

ReportedOct 2026

SWE-MMswe-mm · code

38.6%

ReportedOct 2026

Vision2Webvision2web · code

62.9%

ReportedOct 2026

Instruction Following1 result

IFBenchifbench · instruction_following

79.5%

ReportedOct 2026

Math1 result

MathVisionmathvision · math

94.6%

ReportedOct 2026

Multimodal1 result

OSWorld-Verifiedosworld-verified · multimodal

84.3%

ReportedOct 2026

Reasoning7 results

CharXiv-Rcharxiv-r · reasoning

90.2%

ReportedOct 2026

GPQAgpqa · reasoning

89.2%

ReportedOct 2026

Humanity's Last Examhumanity's-last-exam · reasoning

30.8%

ReportedOct 2026

LiveCodeBench v6livecodebench-v6 · reasoning

90.3%

ReportedOct 2026

SWE-Bench Multimodalswe-bench-multimodal · reasoning

38.6%

ReportedAug 2026

SWE-Bench Proswe-bench-pro · reasoning

61.7%

ReportedOct 2026

Terminal-Bench 2.1terminal-bench-2.1 · reasoning

73.0%

ReportedOct 2026

Vision4 results

BabyVisionbabyvision · vision

85.6%

ReportedOct 2026

ERQAerqa · vision

65.5%

ReportedOct 2026

OmniDocBench 1.5omnidocbench-1.5 · vision

91.1%

ReportedOct 2026

RealWorldQArealworldqa · vision

85.9%

ReportedOct 2026

Learn what each benchmark measures ↗

Pricing detail

Input price per 1M tokens: $0.45
Output price per 1M tokens: $3.20
Blended price per 1M tokens (50/50): $1.83

Prices are published by Venice AI. Venice model page ↗

Capabilities

ReasoningFunction callingVisionVideo inputWeb search

Endpoint metadata

Privacy: private
Quantization: fp8
Maximum completion tokens: 66K
Uncensored: Yes
Beta: Yes
Mapping confidence: 100.0%

View model source ↗

All capabilities

Max Images: 10
Quantization: fp8
Supports E2 E E: No
Supports Vision: Yes
Supports X Search: No
Optimized For Code: Yes
Supports Log Probs: No
Supports Reasoning: Yes
Supports Web Search: Yes
Supports Audio Input: No
Supports Video Input: Yes
Default Reasoning Effort: xhigh
Reasoning Effort Options: none, low, medium, xhigh
Supports Multiple Images: Yes
Supports Response Schema: Yes
Supports Tee Attestation: No
Supports Function Calling: Yes
Supports Reasoning Effort: Yes

Traits

default_vision

Related models

Qwen 3.8 27B

Benchmark results

More from qwen