LLMs TigerLive LLM benchmarks

GLM-5: benchmark scores, price and ranking

GLM-5 from Z.ai (Zhipu AI) ranks #118 of 743 LLMs on the LLMs Tiger aggregate benchmark score (51.5), based on 15 benchmark results captured up to 2026-10-13.

Organisation
Z.ai (Zhipu AI)
Type
Open-weight / local model
Open weights
Yes
Released
2026-02-12 (source: curated)
Cheapest API price
$0.600 in / $1.92 out per million tokens (via openrouter, observed 2026-10-13T19:47Z)
Aggregate rank
#118 of 743 (score 51.5)

Relative to other models, GLM-5 is strongest on MMLU-Pro (#12 of 136) and weakest on ARC-AGI-2 (#59 of 87).

Across the 15 benchmarks where at least ten models are scored, GLM-5 is in the top 10% on 1 and in the top half on 13.

Its cheapest listed input price, $0.600 per million tokens, is lower than that of 44% of the 427 models with a listed price.

GLM-5 benchmark results

GLM-5 benchmark scores, ranked against all models measured on each benchmark
BenchmarkScoreRankSourceCaptured
MMLU-Pro86.0%#12 of 136HuggingFace MMLU-Pro leaderboard2026-10-13
Epoch ECI146#27 of 204Epoch AI Benchmarking Hub2026-10-13
SWE-bench Verified72.8%#14 of 73SWE-bench Verified leaderboard2026-10-13
GPQA Diamond86.0%#71 of 275HuggingFace GPQA Diamond leaderboard2026-10-13
SimpleBench53.2%#16 of 61Epoch AI Benchmarking Hub2026-10-13
WeirdML48.2%#31 of 109Epoch AI Benchmarking Hub2026-10-13
Terminal-Bench 2.052.4%#8 of 28Epoch AI Benchmarking Hub2026-10-13
OTIS Mock AIME80.0%#67 of 189Epoch AI Benchmarking Hub2026-10-13
SWE-bench Multilingual69.7%#4 of 11SWE-bench Multilingual leaderboard2026-10-13
Vending-Bench 2$4,432#8 of 19Epoch AI Benchmarking Hub2026-10-13
CL-bench18.7%#7 of 16Epoch AI Benchmarking Hub2026-10-13
WebDev Arena1,436#32 of 67Epoch AI Benchmarking Hub2026-10-13
ALE-Bench766#39 of 79Epoch AI Benchmarking Hub2026-10-13
Chess Puzzles10.0%#67 of 122Epoch AI Benchmarking Hub2026-10-13
ARC-AGI-24.9%#59 of 87ARC Prize official leaderboard2026-10-13

About GLM-5

GLM-5 is an open-weight model that can run on your own hardware from Z.ai (Zhipu AI). It was released on 2026-02-12, according to the developer's announcement. The cheapest API listing we track charges $0.600 per million input tokens and $1.92 per million output tokens, through openrouter. We hold 15 benchmark results for it across 3 areas, and 1 is in the top tenth of its board.

Where GLM-5 stands

On our aggregate score, GLM-5 is #118 of 743 ranked models, in the top quarter. Its score is 51.5; the scale is centred on 50, so it sits 1.5 points above the typical model. That beats 84% of the ranked field. Among the 6 ranked Z.ai (Zhipu AI) models it is #2. The next one up is GLM-5.3 (max) at 53.8. The next one down is GLM-5.1 at 49.6.

GLM-5, benchmark by benchmark

On MMLU-Pro, GLM-5 scores 86.0% and ranks #12 of 136. Only 11 models score higher, led by Darwin-180B-RSI at 88.1%, which is 2.1 points ahead. It covers 57-subject knowledge + reasoning. The board median is 55.3%, so GLM-5 is 30.7 points above it. Source: HuggingFace MMLU-Pro leaderboard, captured 2026-10-13.

On Epoch ECI, GLM-5 reaches 146. It ranks #27 of 204, ahead of 87% of the field, and 11 points behind Qwen 3.8 Max at 156. The results either side are MiniMax-M2.7 (146) and Gemini 2.5 Pro (Jun 2025) (145). It covers Epoch Capabilities Index: Epoch's own aggregate of ~40 benchmarks. The board median is 127, so GLM-5 is 19 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

GLM-5 manages 72.8% on SWE-bench Verified, good for #14 of 73. Claude Opus 4.7 (max) leads at 83.5%, 10.7 points further up, and 81% of models sit below. The results either side are claude-opus-4-1-20250805 (73.3%) and GPT-5.2 Codex (72.8%). It covers real GitHub issues resolved. The board median is 56.4%, so GLM-5 is 16.4 points above it. Source: SWE-bench Verified leaderboard, captured 2026-10-13.

GPQA Diamond puts GLM-5 at 86.0%, #71 of 275. That beats 74% of the models tested, but 9.8 points short of GPT-6 Astra (max) (95.8%). The results either side are GPT-5 (high) (86.2%) and Qwen3.6 35B-A3B (86.0%). It covers expert-level science questions. The board median is 71.7%, so GLM-5 is 14.3 points above it. Source: HuggingFace GPQA Diamond leaderboard, captured 2026-10-13.

On SimpleBench, GLM-5 reaches 53.2%. It ranks #16 of 61, ahead of 74% of the field, and 28.7 points behind Claude Fable 5 (max) at 81.9%. The results either side are GLM-5.1 (55.1%) and GPT-5.1 (high) (53.2%). It covers trick questions that test common sense. The board median is 41.4%, so GLM-5 is 11.8 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

GLM-5 manages 48.2% on WeirdML, good for #31 of 109. GPT-6 Astra (max) leads at 93.3%, 45.1 points further up, and 72% of models sit below. The results either side are DeepSeek v4 Pro (max) (48.9%) and gpt-oss-120b (high) (48.2%). It covers unusual machine-learning tasks. The board median is 41.2%, so GLM-5 is 7.0 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

WebDev Arena puts GLM-5 at 1,436, #32 of 67. That beats 52% of the models tested, but 384 points short of Claude Opus 5.5 (max) (1,820). The results either side are Kimi K2.5 (1,436) and GLM-4.7 (1,435). It covers building web apps, crowd-voted (Elo). The board median is 1,433, so GLM-5 is 2 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

On ALE-Bench, GLM-5 reaches 766. It ranks #39 of 79, ahead of 51% of the field, and 2,186 points behind GPT-6 Astra (max) at 2,951. The results either side are mimo-v2-pro (785) and Mistral Medium 3.5 (764). It covers AtCoder heuristic-optimisation contests (performance rating). The board median is 764, so GLM-5 is 2 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

GLM-5 lands below the middle on Chess Puzzles: 10.0%, #67 of 122, with 54% of the field scoring higher. GPT-6 Astra (max) leads at 72.0%; the closest result above is Qwen3.5 9B at 12.0%. It covers chess tactics puzzles. The board median is 12.5%, so GLM-5 is 2.5 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

GLM-5 lands below the middle on ARC-AGI-2: 4.9%, #59 of 87, with 67% of the field scoring higher. GPT-6 Astra (max) leads at 95.0%; the closest result above is Gemini 2.5 Pro (Thinking 32K) at 4.9%. It covers abstract reasoning, ARC Prize public eval set. The board median is 22.8%, so GLM-5 is 17.9 points below it. Source: ARC Prize official leaderboard, captured 2026-10-13.

The full table above also lists the other 5 results, including boards with fewer than ten models, where a rank says little.

Strengths and weaknesses

Software engineering: solid, ahead of 64% of the field on average across 4 benchmarks. Its best board here is SWE-bench Verified (#14 of 73) and its weakest is ALE-Bench (#39 of 79). Reasoning, maths and science: solid, ahead of 67% of the field on average across 5 benchmarks. Its best board here is MMLU-Pro (#12 of 136) and its weakest is ARC-AGI-2 (#59 of 87). General knowledge and chat: solid, ahead of 64% of the field on average across 6 benchmarks. Its best board here is Epoch ECI (#27 of 204) and its weakest is Chess Puzzles (#67 of 122).

Price and value

The median input price across the 427 priced models is $0.500 per million tokens, so GLM-5 costs more than the typical listing. Models that score within 1.5 points or better and cost less: Qwen3.7 Flash ($0.030, 50.5), GLM-5.2 (high) ($0.060, 51.6), glm-4.7-flash ($0.060, 50.1). Similar scores cost more on Hy4-preview ($0.751) and claude-3-5-haiku-20241022 ($1.00).

Release history

The Z.ai (Zhipu AI) model released before it was GLM-4.7 (Novita) on 2025-12-22, which scores 48.4. The next one was GLM-5.1 on 2026-04-07, at 49.6.

Reading these numbers

The 15 scores on this page come from 6 sources: Epoch AI Benchmarking Hub (10), ARC Prize official leaderboard (1), HuggingFace GPQA Diamond leaderboard (1), HuggingFace MMLU-Pro leaderboard (1). 2 of them come from an official leaderboard and 13 from datasets or third-party runs.

Models ranked nearby

More from Z.ai (Zhipu AI)

Compare it on the live leaderboard → · How the aggregate score works