LiveCodeBench, Vals AI run (LiveCodeBench (Vals))

We show this table for reference; we do not rank on it.

Data verified September 1, 2026

Vals AI’s independent implementation of the LiveCodeBench code-generation benchmark, run under one fixed harness across the models it tracks.

Evaluation results published by Vals AI. Published leaderboard aggregates and evaluation settings only. Dataset contents are not included.

Vals LiveCodeBench mirror score on LiveCodeBench (Vals) — September 1, 2026

We mirror the published vals livecodebench mirror score view for LiveCodeBench (Vals). Claude Fable 5.1 leads the public snapshot at 90.5%, followed by Claude Fable 5 (89.8%) and Gemini 3.8 Flash (89.5%). We do not use these results to rank models overall.

1Closed

Claude Fable 5.1

Anthropic

90.5%

Overall 81.08Context 1M

2Closed

Claude Fable 5

Anthropic

89.8%

Overall 80.78Context 1M+

3Closed

Gemini 3.8 Flash

Google

high reasoning

89.5%

Overall 66.22Context 1M

143 modelsCoding8% of Coding reference weightCurrentUpdated September 1, 2026

Vals LiveCodeBench mirror score table (143 models)

Vals LiveCodeBench mirror score results for LiveCodeBench (Vals)
Rank	Model / configuration	Score	Parameters (B)	Open / closed
1	Claude Fable 5.1Anthropic	90.5%	Not reported	Closed
2	Claude Fable 5Anthropic	89.8%	Not reported	Closed
3	Gemini 3.8 FlashGooglehigh reasoning	89.5%	Not reported	Closed
4	Claude Opus 5Anthropic	89.0%	Not reported	Closed
5	Gemini 3.7 FlashGooglehigh reasoning	88.7%	Not reported	Closed
6	Gemini 3.1 Pro PreviewGooglehigh reasoning	88.5%	Not reported	Unknown
7	Grok 4.6xAIhigh reasoning	88.2%	Not reported	Closed
8	Gemini 3.6 FlashGooglehigh reasoning	88.1%	Not reported	Closed
9	GPT-5.2-CodexOpenAI	88.0%	Not reported	Closed
10	Qwen3.8 MaxAlibaba	87.9%	Not reported	Open
11	Claude Opus 4.8Anthropic	87.8%	Not reported	Closed
12	Gemini 3.5 FlashGooglehigh reasoning	87.6%	Not reported	Closed
13	DeepSeek V4 Pro 0813DeepSeekmax reasoning	87.5%	Not reported	Open
14	DeepSeek V4 Pro 0813DeepSeekmax reasoning	87.5%	Not reported	Open
15	Grok 4.5xAIhigh reasoning	87.4%	Not reported	Closed
16	GPT-5.3 CodexOpenAIxhigh reasoning	87.3%	Not reported	Closed
17	DeepSeek V4 Flash 0731DeepSeekhigh reasoning	87.3%	Not reported	Open
18	Kimi K3Moonshot AImax reasoning	87.2%	Not reported	Open
19	Qwen3.7 MaxAlibaba	87.1%	Not reported	Closed
20	Kimi K2.6Moonshot AI	86.8%	Not reported	Open
21	GPT-5 miniOpenAIhigh reasoning	86.6%	Not reported	Closed
22	GPT-5.1OpenAIhigh reasoning	86.5%	Not reported	Closed
23	Gemini 3 Pro PreviewGoogle	86.4%	Not reported	Unknown
24	Nemotron 3 Ultra 550b A55bNvidia	86.0%	Not reported	Unknown
25	Qwen3.6 PlusAlibaba	86.0%	Not reported	Closed
26	Inkling-SmallThinking Machines Lab0.99 reasoning	85.9%	Not reported	Open
27	GPT-5.6 TerraOpenAIxhigh reasoning	85.9%	Not reported	Closed
28	GPT-5OpenAIhigh reasoning	85.9%	Not reported	Unknown
29	Muse Spark 1.1Metaxhigh reasoning	85.9%	Not reported	Closed
30	Gemini 3 Flash PreviewGooglehigh reasoning	85.6%	Not reported	Unknown
31	GPT-5.1-CodexOpenAI	85.5%	Not reported	Closed
32	InklingThinking Machines Lab0.99 reasoning	85.5%	Not reported	Open
33	GPT-5.2OpenAIxhigh reasoning	85.4%	Not reported	Closed
34	Qwen3.5 Plus ThinkingAlibaba	85.3%	Not reported	Unknown
35	GPT-5.5OpenAIxhigh reasoning	85.3%	Not reported	Closed
36	Claude Opus 4.7Anthropic	85.1%	Not reported	Closed
37	GPT-5 CodexOpenAIhigh reasoning	84.7%	Not reported	Unknown
38	Claude Opus 4.6 (Adaptive)Anthropic	84.7%	Not reported	Closed
39	Grok 4.3xAIhigh reasoning	84.5%	Not reported	Closed
40	Grok 4.20 0309 ReasoningxAI	84.3%	Not reported	Unknown
41	GPT-5.4OpenAIxhigh reasoning	84.1%	Not reported	Closed
42	GPT-5.4 nanoOpenAIhigh reasoning	84.0%	Not reported	Closed
43	Qwen3.8-27BAlibabaxhigh reasoning	84.0%	Not reported	Open
44	Ling 3.0 Flash 2607Ant	84.0%	Not reported	Unknown
45	o3OpenAIhigh reasoning	83.9%	Not reported	Closed
46	Kimi K2.5 ThinkingMoonshot AI	83.9%	Not reported	Unknown
47	Claude Opus 4.5 ThinkingAnthropic	83.7%	Not reported	Closed
48	GPT-5.1-Codex-MaxOpenAIhigh reasoning	83.6%	Not reported	Closed
49	Qwen3.5 FlashAlibaba	83.3%	Not reported	Closed
50	Grok 4 0709xAI	83.2%	Not reported	Unknown
51	GPT-OSS 120BOpenAI	83.2%	Not reported	Open
52	GPT-5.6 SolOpenAImax reasoning	82.6%	Not reported	Closed
53	Claude Sonnet 5Anthropic	82.4%	Not reported	Closed
54	GLM-4.7Z.AI	82.2%	Not reported	Open
55	o4-miniOpenAIhigh reasoning	82.2%	Not reported	Closed
56	MiniMax M3MiniMax	82.2%	Not reported	Open
57	Claude Sonnet 4.6Anthropic	82.1%	Not reported	Closed
58	Kimi K2.7 CodeMoonshot AI	82.0%	Not reported	Open
59	GLM 5 ThinkingZhipu AI	81.9%	Not reported	Unknown
60	MiniMax M2.1MiniMax	81.8%	Not reported	Unknown
61	MiMo-V2.5Xiaomi	81.5%	Not reported	Closed
62	GPT-5.4 miniOpenAIxhigh reasoning	81.5%	Not reported	Closed
63	GLM-5.1Z.AI	81.4%	Not reported	Open
64	MiMo-V2.5-ProXiaomi	81.4%	Not reported	Closed
65	GLM-4.6Z.AI	81.0%	Not reported	Open
66	DeepSeek V3p2 ThinkingFireworks AI	80.7%	Not reported	Unknown
67	Grok 4.1 Fast (Reasoning)xAI	80.6%	Not reported	Closed
68	GLM-5.3Z.AImax reasoning	80.5%	Not reported	Open
69	GLM-5.3-FlashZ.AImax reasoning	80.5%	Not reported	Open
70	GPT-OSS 20BOpenAI	80.4%	Not reported	Open
71	Gemini 3.1 Flash Lite PreviewGoogle	80.1%	Not reported	Unknown
72	MiniMax M2.7MiniMax	79.9%	Not reported	Open
73	MiniMax M2.5MiniMax	79.2%	Not reported	Closed
74	Gemini 2.5 Pro Preview 03 25Google	79.2%	Not reported	Unknown
75	Gemini 3.5 Flash-LiteGooglehigh reasoning	79.0%	Not reported	Closed
76	Grok 4 Fast (Reasoning)xAI	79.0%	Not reported	Closed
77	Qwen3 MaxAlibaba	78.2%	Not reported	Closed
78	Grok 3 Mini Fast High ReasoningxAIhigh reasoning	76.2%	Not reported	Unknown
79	Gemini 2.5 Flash Preview 09 2025 ThinkingGoogle	76.2%	Not reported	Unknown
80	Gemini 2.5 Flash Preview 09 2025Google	75.1%	Not reported	Unknown
81	Claude Opus 4.5Anthropic	75.0%	Not reported	Closed
82	Magistral Medium 2509Mistral AI	74.9%	Not reported	Unknown
83	Claude Sonnet 4.5 ThinkingAnthropic	73.0%	Not reported	Closed
84	Magistral Small 2509Mistral AI	72.1%	Not reported	Unknown
85	o3-miniOpenAIhigh reasoning	71.5%	Not reported	Closed
86	Gemini 2.5 Flash Lite Preview 09 2025 ThinkingGoogle	71.4%	Not reported	Unknown
87	Qwen3 235B-A22BAlibaba	70.6%	Not reported	Open
88	Moonshotai Kimi K2 InstructTogether AI	70.4%	Not reported	Unknown
89	DeepSeek-R1DeepSeek	70.2%	Not reported	Open
90	GPT-5 nanoOpenAIhigh reasoning	70.2%	Not reported	Closed
91	Claude Opus 4 20250514 ThinkingAnthropic	70.2%	Not reported	Unknown
92	DeepSeek V3p2Fireworks AI	69.9%	Not reported	Unknown
93	GLM-5.2Z.AI	69.5%	Not reported	Open
94	Laguna M.1Poolside	68.1%	Not reported	Closed
95	Laguna XS.2Poolside	67.8%	Not reported	Open
96	Gemini 2.5 Flash Lite Preview 09 2025Google	67.7%	Not reported	Unknown
97	GLM-4.5Z.AI	67.4%	Not reported	Closed
98	Qwen3 Max PreviewAlibaba	66.9%	Not reported	Unknown
99	Claude Opus 4.1 20250805 ThinkingAnthropic	66.5%	Not reported	Unknown
100	Grok 3 Mini Fast Low ReasoningxAIlow reasoning	66.3%	Not reported	Unknown
101	DeepSeek V3 0324DeepSeek	65.5%	Not reported	Open
102	Claude Opus 4.1Anthropic	64.6%	Not reported	Unknown
103	Kimi K2 ThinkingMoonshot AI	63.1%	Not reported	Unknown
104	Claude Opus 4Anthropic	62.6%	Not reported	Unknown
105	Claude Sonnet 4 20250514 ThinkingAnthropic	62.4%	Not reported	Unknown
106	Grok Code Fast 1xAI	62.0%	Not reported	Closed
107	Claude 3.7 Sonnet ThinkingAnthropic	60.4%	Not reported	Closed
108	Claude Sonnet 4Anthropic	59.7%	Not reported	Unknown
109	Langston Nim Nvidia Llama 3.3 Nemotron Super 49b V1 42e84561 ThinkingTogether AI	58.4%	Not reported	Unknown
110	GPT-4.1 miniOpenAIhigh reasoning	58.2%	Not reported	Closed
111	Gemini 2.5 Flash Preview 04 17Google	56.9%	Not reported	Unknown
112	Claude 3.7 SonnetAnthropic	56.7%	Not reported	Closed
113	Mistral Large 2512Mistral AI	55.3%	Not reported	Unknown
114	GPT-4.1OpenAIhigh reasoning	54.7%	Not reported	Closed
115	Grok 3xAI	52.9%	Not reported	Unknown
116	Devstral 2512Mistral AI	51.8%	Not reported	Unknown
117	o1OpenAIhigh reasoning	50.3%	Not reported	Closed
118	Claude 3.5 SonnetAnthropic	49.6%	Not reported	Closed
119	Llama4 Maverick Instruct BasicFireworks AI	47.3%	Not reported	Unknown
120	Gemini 2.5 Flash Preview 04 17 ThinkingGoogle	46.9%	Not reported	Unknown
121	Grok 4 Fast Non ReasoningxAI	46.1%	Not reported	Unknown
122	Mistral Medium 2505Mistral AI	44.8%	Not reported	Unknown
123	Gemini 2.0 Flash 001Google	43.6%	Not reported	Unknown
124	GPT-4oOpenAIhigh reasoning	43.4%	Not reported	Closed
125	Labs Devstral Small 2512Mistral AI	43.2%	Not reported	Unknown
126	GPT-4.1 nanoOpenAIhigh reasoning	42.7%	Not reported	Closed
127	Grok 4.1 Fast Non ReasoningxAI	42.6%	Not reported	Unknown
128	Claude 3.5 HaikuAnthropic	41.9%	Not reported	Closed
129	Gemini 1.5 Pro 002Google	41.7%	Not reported	Unknown
130	Claude Haiku 4.5 ThinkingAnthropic	41.2%	Not reported	Closed
131	Grok 2 1212xAI	38.7%	Not reported	Unknown
132	Meta Llama Llama 4 Scout 17B 16E InstructTogether AI	38.5%	Not reported	Unknown
133	Mistral Large 2411Mistral AI	37.1%	Not reported	Unknown
134	Gemini 1.5 Flash 002Google	36.9%	Not reported	Unknown
135	Meta Llama Llama 3.3 70B Instruct TurboTogether AI	36.3%	Not reported	Unknown
136	Langston Nim Nvidia Llama 3.3 Nemotron Super 49b V1 42e84561Together AI	36.3%	Not reported	Unknown
137	Command A 03 2025Cohere	35.1%	Not reported	Unknown
138	Mistral Small 2503Mistral AI	31.8%	Not reported	Unknown
139	GPT-4o miniOpenAIhigh reasoning	26.4%	Not reported	Closed
140	Jamba Large 1.6AI21 Labs	22.3%	Not reported	Unknown
141	Command R PlusCohere	18.2%	Not reported	Unknown
142	Mistral Small 2402Mistral AI	15.8%	Not reported	Unknown
143	Jamba Mini 1.6AI21 Labs	9.9%	Not reported	Unknown

How Vals LiveCodeBench mirror is shown here

BenchLM mirrors the public Vals AI Vals LiveCodeBench mirror leaderboard captured from https://www.vals.ai/benchmarks/lcb and updated by Vals on September 1, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.

Each row keeps the harness and settings the source published. BenchLM shows the published table for reference; the per-model scores it stores feed the ranking.

Snapshot

143 Vals rows4 task viewspublic datasetTasks: Overall, Easy, Medium, HardScored

Vals LiveCodeBench mirror on Vals AI Vals methodology Vals home

The published LiveCodeBench (Vals) snapshot places Claude Fable 5.1 first at 90.5%. The third row is 1.0 points behind. The broader top-10 range is 2.7 points, so many of the published results sit in a relatively narrow band.

143 models have been evaluated on LiveCodeBench (Vals). The benchmark falls in the Coding category. BenchLM shows the published table for reference; the per-model scores it stores feed the ranking. BenchAlign v5.8 gives LiveCodeBench (Vals) 8% of the Coding reference weight, so it moves the Coding leaderboard and the overall ranking. Reference weights are relative weights in the calibrated model, not fixed shares of a score.

About LiveCodeBench (Vals)

Year

2026

Tasks

Competitive programming problems (easy, medium, hard)

Format

Pass@1 accuracy

Difficulty

Frontier coding

BenchLM mirrors the Vals AI board on a dedicated key so a provider-run row on the canonical key is never overwritten. Vals publishes per-task accuracy with standard error, latency, and cost for every model it runs. Admitted as independent third-party evidence in methodology v5.5 (2026-09-04).

Vals AI LiveCodeBench, Vals AI run leaderboard Public benchmark source

Freshness and provenance

Version

LiveCodeBench (Vals) 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

Current

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

Questions

What does LiveCodeBench (Vals) measure?

Vals AI’s independent implementation of the LiveCodeBench code-generation benchmark, run under one fixed harness across the models it tracks.

Which model leads the published LiveCodeBench (Vals) snapshot?

Claude Fable 5.1 currently leads the published LiveCodeBench (Vals) snapshot with 90.5% vals livecodebench mirror score. BenchLM shows the published table for reference; the per-model scores it stores feed the ranking.

How many models are evaluated on LiveCodeBench (Vals)?

The September 1, 2026 snapshot contains 143 AI models.

Last updated: September 1, 2026 · mirrored from the public benchmark leaderboard

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 6,100+ subscribers.

One email each week. Unsubscribe anytime.