# Sources (accessed 2026-07-20):
# https://docs.qwencloud.com/token-plan/personal/token-plan-personal-overview
# https://platform.qianwenai.com/docs/token-plan/personal/token-plan-personal-overview
# https://docs.qwencloud.com/developer-guides/getting-started/text-generation-models
# https://platform.qianwenai.com/docs/developer-guides/getting-started/text-generation-models
# https://docs.qwencloud.com/developer-guides/clients-and-developer-tools/opencode
# https://platform.qianwenai.com/docs/developer-guides/clients-and-developer-tools/opencode
# https://docs.qwencloud.com/developer-guides/clients-and-developer-tools/kilo-cli
# https://platform.qianwenai.com/docs/developer-guides/clients-and-developer-tools/kilo-cli
# https://github.com/QwenLM/qwen-code/issues/7198
# https://github.com/QwenLM/qwen-code/pull/7199

name = "Qwen3.8 Max Preview"
description = "Preview Qwen flagship for million-token multimodal reasoning and long-horizon agentic workflows"
family = "qwen"
release_date = "2026-07-19"
last_updated = "2026-07-19"
attachment = true
reasoning = true
temperature = true
tool_call = true
open_weights = false

[limit]
context = 1_000_000
output = 131_072

[modalities]
input = ["text", "image", "video"]
output = ["text"]

[[benchmarks]]
name = "Terminal-Bench"
score = 86.6
metric = "accuracy"
variant = "xhigh"
version = "2.1"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "SWE-Bench Pro"
score = 67.7
metric = "resolve rate"
variant = "xhigh"
harness = "Claude Code"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "DeepSWE"
score = 56.6
metric = "resolve rate"
variant = "xhigh"
harness = "Claude Code"
version = "1.1"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "NL2Repo"
score = 55.9
metric = "resolve rate"
variant = "xhigh"
harness = "Claude Code"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "FrontierSWE"
score = 73.5
metric = "dominance score"
variant = "xhigh"
harness = "Claude Code"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "MLS-Bench-Lite"
score = 41.0
metric = "score"
variant = "xhigh"
harness = "Claude Code"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "AutomationBench"
score = 27.3
metric = "pass@1"
variant = "xhigh"
dataset = "600-task public subset"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "Toolathlon Verified"
score = 72.5
metric = "pass@1"
variant = "xhigh"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "WideSearch"
score = 81.9
metric = "F1"
variant = "xhigh"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "Humanity's Last Exam"
score = 56.2
metric = "accuracy"
variant = "xhigh, with tools"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "GPQA Diamond"
score = 92.6
metric = "accuracy"
variant = "xhigh"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "Humanity's Last Exam"
score = 43.6
metric = "accuracy"
variant = "xhigh, no tools"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "IFBench"
score = 82.8
metric = "score"
variant = "xhigh"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "OSWorld-Verified"
score = 86.1
metric = "success rate"
variant = "xhigh"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"

[[benchmarks]]
name = "MMMU Pro"
score = 82.3
metric = "accuracy"
variant = "xhigh"
source = "https://www.alibabacloud.com/blog/qwen3-8-max-a-new-bar-for-coding-and-cowork_603421"
date = "2026-08-03"