Open models, explained.

Which open model is best for your use case, what it costs, and how it compares to closed models.

Pick by use case

One index hides the fit. Choose what you are building; the answer and the evidence follow.

Intelligence Coding Agents Long context Multilingual VisionEngineering Legal Healthcare Finance Economics Strategy

Which model is smartest overall?

71 models scored · AA Intelligence Index, with 12 measures beneath it · Artificial Analysis

Open pick

ZGLM-5.345

$2.01 per task · 84% of the leader's score for 81% less

Best value

ZGLM-5.3-Flash42

$0.25 per task · 93% of the open pick, 11% of the cost

To beat

Claude Fable 5.153

$7.63 per task · 8.5 points ahead of the open pick

Open sourceClosed

Score breakdownAA-Briefcase · GDPval-AA v2 · Terminal-Bench 4.0 · SciCode · Humanity's Last Exam · CritPt · GDP.pdf · Omniscience accuracy · Not guessing · AutomationBench-AA · AA-LCR v1.1 · GPQA Diamond

Model	Score	Cost / task	AA-Briefcase15%	GDPval-AA v210%	Terminal-Bench 4.010%	SciCode10%	Humanity's Last Exam10%	CritPt10%	GDP.pdf10%	Omniscience accuracy10%	Not guessing5%	AutomationBench-AA5%	AA-LCR v1.15%	GPQA Diamond
Claude Fable 5.1	53.4	$7.63	58.1	63.2	52.0	63.1	59.1	29.7	26.2	67.2	27.4	59.4	85.3	93.7
GPT-6 Astra	52.8	$3.26	53.1	54.0	59.1	56.5	54.7	31.7	31.0	62.6	48.7	68.5	80.7	96.1
Claude Opus 5	50.7	$5.86	57.2	61.8	49.0	56.4	54.9	29.1	21.6	60.9	39.2	56.6	79.3	93.2
Claude Fable 5	49.7	$8.75	51.5	56.6	42.4	61.0	55.5	28.6	24.0	65.3	36.4	54.1	82.3	92.6
Muse Spark 1.3	48.2	$1.60	54.5	60.2	33.3	58.8	48.7	24.9	26.6	43.6	67.1	57.9	83.0	93.5
GPT-5.6 Sol	47.1	$1.99	48.8	56.2	39.9	57.1	49.5	32.3	27.2	59.4	7.8	60.1	84.0	94.1
Qwen3.8 Max	45.4	$5.41	56.1	59.4	38.9	52.1	43.1	17.7	22.8	31.7	71.2	56.2	80.3	92.8
ZGLM-5.3	44.9	$2.01	50.6	57.8	41.9	59.0	42.3	19.1	11.2	33.9	70.4	62.2	79.7	91.7
Grok 4.6	44.4	$1.86	51.7	57.1	21.2	56.5	42.9	17.1	17.0	48.2	65.7	66.7	80.3	94.9
Kimi K3	43.8	$2.00	49.4	52.6	12.6	59.5	46.9	23.4	22.0	47.6	46.8	58.3	88.7	93.5

The score is AA Intelligence Index exactly as Artificial Analysis publishes it; the columns are what goes into it. Best result per column is underlined. A “·” means Artificial Analysis has not run that measure against the model; it is left as a gap, never counted as a zero. † is still measured by AA but no longer counted by Intelligence Index v4.3, so it is shown for context and not averaged into the score.

Every model, ranked

The whole field on one index. Sort, filter, click to inspect.

Artificial Analysis Intelligence Index

Composite of ten public evals: agents, coding, scientific reasoning, general · Higher is better

Open sourceClosed

Click a bar to inspect

top 18 of 71 shown · full list belowAttributedArtificial Analysis v4.3

Leaderboard

	Model							License
1	Claude Fable 5.1 Anthropic	53	0.4	$13.1K	$10.0 / $50.0	65 t/s	1M	Proprietary
2	GPT-6 Astra OpenAI	53	1.0	$5.3K	$10.0 / $50.0	52 t/s	1M	Proprietary
3	Claude Opus 5 Anthropic	51	0.7	$7.3K	$5.00 / $25.0	50 t/s	1M	Proprietary
4	Claude Fable 5 Anthropic	50	0.4	$11.2K	$10.0 / $50.0	67 t/s	1M	Proprietary
5	Muse Spark 1.3 Meta	48	2.4	$2.0K	$1.25 / $4.25	235 t/s	1M	Proprietary
6	GPT-5.6 Sol OpenAI	47	1.4	$3.5K	$4.00 / $20.0	67 t/s	1M	Proprietary
7	Qwen3.8 Max Alibaba (Qwen)	45	0.9	$4.9K	$2.00 / $6.00	40 t/s	984K	Proprietary
8	Z GLM-5.3 Z.ai (GLM)	45	1.8	$2.5K	$1.40 / $4.40	228 t/s	1M	GLM-5.3 License
9	Grok 4.6 xAI	44	1.9	$2.4K	$2.00 / $6.00	63 t/s	500K	Proprietary
10	Kimi K3 Moonshot (Kimi)	44	1.2	$3.7K	$3.00 / $15.0	60 t/s	1.0M	Kimi K3 License

GLM-5.3

Z.ai (GLM) · GLM-5.3 License · 753B · 40B active

open

Intelligence Index45

Intel / $1001.8

Cost per task: $2.01
Full index run: $2.5K
Price in / out: $1.40 / $4.40
Cached input: 5x cheaper$0.26
Speed: 0.60s TTFT228 t/s
Context: 1M

Capability breakdown0 to 100

Agents

Coding

Reasoning

Knowledge

Long context

zai-org/GLM-5.3Run it

The state of open, at a glance

Who is smartest, and who gives the most per dollar.

Open sourceClosed

Intelligence

Artificial Analysis Intelligence Index · Higher is better

Value

Index points per $100 of run cost · models scoring 32+

GLM-5.3 is 8.5 points behind Claude Fable 5.1 on the Intelligence Index, holding 84% of the top score. In 2025 that gap was 9.3 points, and it has narrowed since.

GLM-5.3

Open source · 45

Claude Fable 5.1

Closed · 53

Run the open frontier yourself

Same OpenAI-compatible API, open source weights you keep, at a fraction of closed-API cost. The snippet is pointed at GLM-5.3, the highest-scoring open model Together serves today and the open leader overall.

Start on Together What would it cost?

Copied

# uv add together

from together import Together

client = Together()

response = client.chat.completions.create(

model="zai-org/GLM-5.3",

messages=[{"role": "user", "content": "What are the top 3 things to do in New York?"}],

)

print(response.choices[0].message.content)