llmboard.aiAI model intelligence
Home

Model Rankings

OverallOpen ModelsAgentCodingReasoningMathKnowledgeInstruction FollowingTextVision
Image GenerationImage Editing
Video GenerationImage to VideoVideo Editing
Text to SpeechSpeech to Text
Embeddings

Efficiency

Chat Token PricingImage PricingVideo PricingAudio Pricing
Chat Speed & LatencyProvider Reliability

Benchmarks

GPQAMMLU-ProAIME 2025SWE-Bench VerifiedMMLUHumanity's Last ExamLiveCodeBenchMATHHumanEvalMMMU-Pro
All Benchmarks

Tools

Model DirectoryCompare Models

Scoring & Data

Scoring & Data
393 models668 benchmarks

Leaderboard Center

Overall RankingCodingCore BenchmarksPrice & ValueRuntime Performance

Modalities

All ModelsImage GenerationImage EditingVideo GenerationImage-to-VideoVideo EditingText-to-SpeechSpeech-to-TextEmbeddings

Data & Methods

Scoring MethodAll BenchmarksReasoningMath

Vendors

All VendorsOpenAIAnthropicGoogle
llmboard.aiCopyright 2026 llmboard.ai

Open weights

Open Model Leaderboard

Use this open model leaderboard to compare the current open-weight ranking from aggregated benchmark evidence, capability, context, licenses, pricing and runtime.

Data as of 2026-08-17

On this page

  • Ranking
  • Overview
  • Capability
  • Landscape
  • Value
  • Compare
  • Top models
  • FAQ

Open Model Leaderboard Ranking

This leaderboard ranks each model using its latest scored open-weight version and eligible benchmark evidence.

30 of 70 rows
Columns

Show columns

Sort by
Rank
Model
LLMBoard Score
License
Parameters
Context
Official input / 1M
Output speed
Rank01ModelMAKimi K3Moonshot AILLMBoard Score92.0LicenseOpen weightsParameters2.8TContext1MOfficial input / 1M$3Output speed25.734 tok/s
Rank02ModelZAGLM 5.2Zhipu AILLMBoard Score79.5LicenseOpen weightsParameters753BContext1MOfficial input / 1M$1.4Output speedN/A
Rank03ModelACQwen3.8 27BAlibaba Cloud / Qwen TeamLLMBoard Score77.6LicenseOpen weightsParameters27.8BContext262.1KOfficial input / 1MN/AOutput speedN/A
Rank04ModelDEDeepSeek-V4-FlashDeepSeekLLMBoard Score77.4LicenseOpen weightsParameters304BContext1MOfficial input / 1M$0.14Output speed19.447 tok/s
Rank05ModelMAKimi K2.6Moonshot AILLMBoard Score75.1LicenseOpen weightsParameters1TContext262.1KOfficial input / 1M$0.95Output speed285.051 tok/s
Rank06ModelMIMiniMax M3MiniMaxLLMBoard Score71.1LicenseOpen weightsParameters428BContext512KOfficial input / 1M$0.30Output speed213.561 tok/s
Rank07ModelTEHy3TencentLLMBoard Score70.7LicenseOpen weightsParameters295BContext256KOfficial input / 1MN/AOutput speedN/A
Rank08ModelPOLaguna S 2.1PoolsideLLMBoard Score69.8LicenseOpen weightsParameters118BContext1MOfficial input / 1MN/AOutput speedN/A
Rank09ModelZAGLM 5.1Zhipu AILLMBoard Score69.3LicenseOpen weightsParameters754BContext200KOfficial input / 1M$1.4Output speedN/A
Rank10ModelZAGLM 5Zhipu AILLMBoard Score67.1LicenseOpen weightsParameters744BContext204.8KOfficial input / 1M$1Output speed30 tok/s
Rank11ModelMAKimi K2.7 CodeMoonshot AILLMBoard Score67.0LicenseOpen weightsParameters1TContext262.1KOfficial input / 1M$0.95Output speed36.107 tok/s
Rank12ModelACQwen3.5 397B A17BAlibaba Cloud / Qwen TeamLLMBoard Score65.9LicenseOpen weightsParameters397BContext262.1KOfficial input / 1M$0.60Output speedN/A
Rank13ModelMAKimi K2.5Moonshot AILLMBoard Score65.7LicenseOpen weightsParameters1TContext262.1KOfficial input / 1M$0.60Output speedN/A
Rank14ModelTMInkling SmallThinking Machines LabLLMBoard Score65.6LicenseOpen weightsParameters276BContext1MOfficial input / 1MN/AOutput speedN/A
Rank15ModelNVNemotron 3 UltraNVIDIALLMBoard Score62.7LicenseOpen weightsParameters550BContext1MOfficial input / 1M$0.50Output speedN/A
Rank16ModelXIMiMoXiaomiLLMBoard Score61.6LicenseOpen weightsParameters310.8BContext1MOfficial input / 1M$0.14Output speed87.039 tok/s
Rank17ModelSTStep 3.5 FlashStepFunLLMBoard Score60.3LicenseOpen weightsParameters196BContext256KOfficial input / 1M$0.10Output speed711.842 tok/s
Rank18ModelACQwen3.6 27BAlibaba Cloud / Qwen TeamLLMBoard Score59.9LicenseOpen weightsParameters27.8BContext262.1KOfficial input / 1M$0.60Output speed6.118 tok/s
Rank19ModelGOGemma 4 31BGoogleLLMBoard Score59.7LicenseOpen weightsParameters30.7BContext262.1KOfficial input / 1MN/AOutput speed8.296 tok/s
Rank20ModelZAGLM 4.7Zhipu AILLMBoard Score59.4LicenseOpen weightsParameters358BContext204.8KOfficial input / 1M$0.60Output speedN/A
Rank21ModelMIMiniMax M2.7MiniMaxLLMBoard Score59.1LicenseOpen weightsParametersN/AContext204.8KOfficial input / 1M$0.30Output speed101.856 tok/s
Rank22ModelACQwen3.5 122B A10BAlibaba Cloud / Qwen TeamLLMBoard Score58.9LicenseOpen weightsParameters122BContext262.1KOfficial input / 1M$0.40Output speedN/A
Rank23ModelMEMuse Glimmer 30BMetaLLMBoard Score58.5LicenseOpen weightsParameters29.6BContext131.1KOfficial input / 1MN/AOutput speedN/A
Rank24ModelACQwen3.5 27BAlibaba Cloud / Qwen TeamLLMBoard Score56.8LicenseOpen weightsParameters27BContext262.1KOfficial input / 1M$0.30Output speed6.736 tok/s
Rank25ModelMIMiniMax M2.5MiniMaxLLMBoard Score55.7LicenseOpen weightsParameters230BContext204.8KOfficial input / 1M$0.30Output speed100 tok/s
Rank26ModelACQwen3.6 35B A3BAlibaba Cloud / Qwen TeamLLMBoard Score54.8LicenseOpen weightsParameters35BContext262.1KOfficial input / 1M$0.248Output speedN/A
Rank27ModelZAGLM 4.6Zhipu AILLMBoard Score53.6LicenseOpen weightsParameters357BContext204.8KOfficial input / 1M$0.60Output speed85 tok/s
Rank28ModelMIMiniMax M2.1MiniMaxLLMBoard Score53.3LicenseOpen weightsParameters230BContext204.8KOfficial input / 1M$0.30Output speed100 tok/s
Rank29ModelGOGemma 4 26B A4BGoogleLLMBoard Score52.2LicenseOpen weightsParameters25.2BContext262.1KOfficial input / 1MN/AOutput speedN/A
Rank30ModelACQwen3.5 35B A3BAlibaba Cloud / Qwen TeamLLMBoard Score51.6LicenseOpen weightsParameters35BContext262.1KOfficial input / 1M$0.25Output speedN/A
Rank31ModelXIMiMo V2 FlashXiaomiLLMBoard Score51.0LicenseOpen weightsParameters309BContext262.1KOfficial input / 1M$0.14Output speedN/A
Rank32ModelZAGLM 4.5Zhipu AILLMBoard Score49.1LicenseOpen weightsParameters355BContext131.1KOfficial input / 1M$0.60Output speedN/A
Rank33ModelMAMistral Medium 3.5Mistral AILLMBoard Score48.7LicenseOpen weightsParameters128BContext262.1KOfficial input / 1M$1.5Output speed0.961 tok/s
Rank34ModelXIMiMo V2.5 ProXiaomiLLMBoard Score46.2LicenseOpen weightsParameters1TContext1MOfficial input / 1M$0.435Output speed68.716 tok/s
Rank35ModelMIMiniMax M2MiniMaxLLMBoard Score44.3LicenseOpen weightsParameters230BContext196.6KOfficial input / 1M$0.30Output speed103.094 tok/s
Rank36ModelNVNemotron 3 SuperNVIDIALLMBoard Score44.3LicenseOpen weightsParameters120BContext262.1KOfficial input / 1M$0.20Output speedN/A
Rank37ModelSASarvam 105BSarvam AILLMBoard Score43.8LicenseOpen weightsParameters105BContext131.1KOfficial input / 1MN/AOutput speedN/A
Rank38ModelACQwen3 235B A22BAlibaba Cloud / Qwen TeamLLMBoard Score41.9LicenseOpen weightsParameters235BContext262.1KOfficial input / 1M$0.70Output speed68.17 tok/s
Rank39ModelZAGLM 4.5 AirZhipu AILLMBoard Score41.5LicenseOpen weightsParameters106BContext131.1KOfficial input / 1M$0.20Output speedN/A
Rank40ModelCOCommand A+CohereLLMBoard Score41.4LicenseOpen weightsParameters218BContext128KOfficial input / 1M$2.5Output speedN/A
Rank41ModelPOLaguna XS 2.1PoolsideLLMBoard Score40.9LicenseOpen weightsParameters33BContext262.1KOfficial input / 1MN/AOutput speedN/A
Rank42ModelACQwen3.5 9BAlibaba Cloud / Qwen TeamLLMBoard Score40.9LicenseOpen weightsParameters9BContext262.1KOfficial input / 1MN/AOutput speedN/A
Rank43ModelACQwen3 Next 80B A3B ThinkingAlibaba Cloud / Qwen TeamLLMBoard Score40.4LicenseOpen weightsParameters80BContext131.1KOfficial input / 1M$0.50Output speedN/A
Rank44ModelDEDeepSeek-V3.1DeepSeekLLMBoard Score39.2LicenseOpen weightsParameters671BContext131.1KOfficial input / 1MN/AOutput speedN/A
Rank45ModelZAGLM 4.7 FlashZhipu AILLMBoard Score37.9LicenseOpen weightsParameters30BContext200KOfficial input / 1MN/AOutput speed50 tok/s
Rank46ModelNVNemotron 3.5 LightningNVIDIALLMBoard Score37.3LicenseOpen weightsParameters30BContext262.1KOfficial input / 1MN/AOutput speedN/A
Rank47ModelACQwen3 Next 80B A3BAlibaba Cloud / Qwen TeamLLMBoard Score35.9LicenseOpen weightsParameters80BContext131.1KOfficial input / 1M$0.50Output speedN/A
Rank48ModelACQwen3 Coder 480B A35BAlibaba Cloud / Qwen TeamLLMBoard Score35.0LicenseOpen weightsParameters480BContext262.1KOfficial input / 1M$1.5Output speedN/A
Rank49ModelNVNemotron 3 NanoNVIDIALLMBoard Score34.9LicenseOpen weightsParameters32BContext262.1KOfficial input / 1MN/AOutput speedN/A
Rank50ModelSASarvam 30BSarvam AILLMBoard Score33.5LicenseOpen weightsParameters30BContext128KOfficial input / 1MN/AOutput speedN/A
Rank51ModelMAMistral Small 4Mistral AILLMBoard Score32.6LicenseOpen weightsParameters119BContext256KOfficial input / 1M$0.15Output speed4.301 tok/s
Rank52ModelNVNemotron Nano 9BNVIDIALLMBoard Score31.7LicenseOpen weightsParameters8.9BContext131.1KOfficial input / 1MN/AOutput speedN/A
Rank53ModelACQwen3 32BAlibaba Cloud / Qwen TeamLLMBoard Score30.8LicenseOpen weightsParameters32.8BContext131.1KOfficial input / 1M$0.70Output speed327.7 tok/s
Rank54ModelACQwen3 30B A3BAlibaba Cloud / Qwen TeamLLMBoard Score28.2LicenseOpen weightsParameters30.5BContext131.1KOfficial input / 1MN/AOutput speed122.4 tok/s
Rank55ModelDEDeepSeek-V3DeepSeekLLMBoard Score25.5LicenseOpen weightsParameters671BContext163.8KOfficial input / 1MN/AOutput speed100 tok/s
Rank56ModelACQwQ 32BAlibaba Cloud / Qwen TeamLLMBoard Score24.2LicenseOpen weightsParameters32.5BContext131.1KOfficial input / 1M$0.287Output speed99.15 tok/s
Rank57ModelGOGemma 4 E4BGoogleLLMBoard Score23.7LicenseOpen weightsParameters8BContext131.1KOfficial input / 1MN/AOutput speedN/A
Rank58ModelNVLlama 3.3 Nemotron Super 49BNVIDIALLMBoard Score22.3LicenseOpen weightsParameters49.9BContext131.1KOfficial input / 1MN/AOutput speedN/A
Rank59ModelACQwen2.5 VL 72BAlibaba Cloud / Qwen TeamLLMBoard Score22.2LicenseOpen weightsParameters72BContext131.1KOfficial input / 1MN/AOutput speedN/A
Rank60ModelMELlama 3.3 70BMetaLLMBoard Score21.4LicenseOpen weightsParameters70BContext128KOfficial input / 1MN/AOutput speed2,220 tok/s
Rank61ModelMAMagistral SmallMistral AILLMBoard Score19.3LicenseOpen weightsParameters24BContext131.1KOfficial input / 1MN/AOutput speedN/A
Rank62ModelMADevstral Small 1.1Mistral AILLMBoard Score16.5LicenseOpen weightsParameters24BContext128KOfficial input / 1M$0.10Output speed137.1 tok/s
Rank63ModelMAMistral Large 3Mistral AILLMBoard Score15.3LicenseOpen weightsParameters675BContext262.1KOfficial input / 1M$0.50Output speed0.778 tok/s
Rank64ModelGOGemma 4 E2BGoogleLLMBoard Score11.0LicenseOpen weightsParameters5.1BContext131.1KOfficial input / 1MN/AOutput speedN/A
Rank65ModelACQwen2.5 Coder 32BAlibaba Cloud / Qwen TeamLLMBoard Score4.9LicenseOpen weightsParameters32BContext131.1KOfficial input / 1MN/AOutput speed110 tok/s
Rank66ModelNVLlama 3.1 Nemotron 70BNVIDIALLMBoard Score2.6LicenseOpen weightsParameters70BContext128KOfficial input / 1MN/AOutput speedN/A
Rank67ModelMAPixtral 12BMistral AILLMBoard Score0.0LicenseOpen weightsParameters12.4BContext128KOfficial input / 1M$0.15Output speed0.1 tok/s
Rank68ModelMAMinistral 8BMistral AILLMBoard Score0.0LicenseOpen weightsParameters8BContext131.1KOfficial input / 1MN/AOutput speed0.1 tok/s
Rank69ModelMELlama 3.1 8BMetaLLMBoard Score0.0LicenseOpen weightsParameters8BContext128KOfficial input / 1MN/AOutput speed2,047 tok/s
Rank70ModelMIPhi 4 MiniMicrosoftLLMBoard Score0.0LicenseOpen weightsParameters3.8BContext128KOfficial input / 1M$0.075Output speedN/A

Open-weight status belongs to the scored model version. License terms still need to be checked on the model detail page.

Open Model Leaderboard Overview

Key leaders and coverage signals from the current open-weight ranking.

Top open-weight modelKimi K392.0 LLMBoard scoreLargest contextKimi K31M tokensMost recent releaseQwen3.8 27B2026-08-14
Ranked open models70Unique model products

Open Model Benchmark Leaders

Each capability benchmark remains a separate score alongside the overall leaderboard.

Reasoning97.5Kimi K3Rank #1 overall
Knowledge89.8Kimi K3Rank #1 overall
Math100.0MiniMax M3Rank #6 overall
Coding96.9Llama 3.3 Nemotron Super 49BRank #58 overall
Instruction97.0Inkling SmallRank #14 overall

Open-model leaderboard landscape

Vendor, license and model-size patterns across models in the current ranking.

Top-30 vendor mix

30 models
Alibaba Cloud / Qwen Team7Zhipu AI5Moonshot AI4MiniMax4Google2DeepSeek1Tencent1Poolside1

Scored-version licenses

Apache 2.0
32
MIT
20
OpenMDW License v1.1
4
Modified MIT License
3
NVIDIA Open Model License Agreement
3
Llama 3.1 Community License
3
Kimi K3 License
1
02550751002.3B15.3B103.7B701.4B4.7TLLMBoard scoreParameters
69 models with comparable dataOpen a model by selecting its logo

Official price and capability

Only positive official vendor PAYG prices are included. Price and benchmark-backed capability remain separate leaderboard signals.

020406080100$0.1$0.2$0.5$1.0$2.0$5.0LLMBoard scoreOfficial API price blend (8:1 input/output), USD / 1M
Efficient frontier42 models with official PAYG prices

Compare leading open models

#2-12.5
ZA

GLM 5.2

Zhipu AI

79.5 LLMBoard

DetailsCompare
#3-14.4
AC

Qwen3.8 27B

Alibaba Cloud / Qwen Team

77.6 LLMBoard

DetailsCompare
#4-14.6
DE

DeepSeek-V4-Flash

DeepSeek

77.4 LLMBoard

DetailsCompare
#5-16.9
MA

Kimi K2.6

Moonshot AI

75.1 LLMBoard

DetailsCompare
#6-21.0
MI

MiniMax M3

MiniMax

71.1 LLMBoard

DetailsCompare
#7-21.4
TE

Hy3

Tencent

70.7 LLMBoard

DetailsCompare

The Top AI Models for Open-Weight Use

The current top five entries in the open model leaderboard, summarized with benchmark capability, official price, output speed and deployment-relevant evidence.

Ranking basisThis open-weight use AI model leaderboard uses the current LLMBoard capability score among versions marked as open weight. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.

  1. 01
    MA
    Kimi K3Moonshot AI
    LLMBoard Score
    92.0
    Price
    $3.0 input / $15 output per 1M tokens
    Speed
    Up to 26 tok/s via Fireworks

    Strengths

    • Ranks #1 with 92.0 LLMBoard Score
    • The scored version is marked as open weight
    • 1M-token context window

    Considerations

    • Evidence coverage is 60%
  2. 02
    ZA
    GLM 5.2Zhipu AI
    LLMBoard Score
    79.5
    Price
    $1.4 input / $4.4 output per 1M tokens

    Strengths

    • Ranks #2 with 79.5 LLMBoard Score
    • The scored version is marked as open weight
    • 1M-token context window
  3. 03
    AC
    Qwen3.8 27BAlibaba Cloud / Qwen Team
    LLMBoard Score
    77.6

    Strengths

    • Ranks #3 with 77.6 LLMBoard Score
    • The scored version is marked as open weight
    • 262.1K-token context window
  4. 04
    DE
    DeepSeek-V4-FlashDeepSeek
    LLMBoard Score
    77.4
    Price
    $0.14 input / $0.28 output per 1M tokens
    Speed
    Up to 19 tok/s via Novita

    Strengths

    • Ranks #4 with 77.4 LLMBoard Score
    • The scored version is marked as open weight
    • Lowest official input price among the current top five

    Considerations

    • Provisional result with 20% evidence coverage
  5. 05
    MA
    Kimi K2.6Moonshot AI
    LLMBoard Score
    75.1
    Price
    $0.95 input / $4.0 output per 1M tokens
    Speed
    Up to 285 tok/s via Fireworks

    Strengths

    • Ranks #5 with 75.1 LLMBoard Score
    • The scored version is marked as open weight
    • 262.1K-token context window

Selection summary

Best AI Models for Open-Weight Use

Kimi K3 is the highest-ranked open-weight model at 92.0. The best open LLM also depends on license terms, hardware needs, context, price and available runtime records.

Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.

Open-model rank #1Kimi K3Up to 26 tok/s via FireworksOpen-model rank #2GLM 5.2$1.4 input / $4.4 output per 1M tokensOpen-model rank #3Qwen3.8 27B77.6

About open-weight models

Common questions about this open model leaderboard, benchmark coverage and ranking.

What is the highest-ranked open-weight model?

Kimi K3 currently leads the open-model leaderboard ranking.

What does open weights mean here?

The scored model version is marked as having publicly available weights. That flag does not by itself determine whether the license permits every commercial or derivative use.

How many models are ranked?

This page currently ranks 70 unique open-weight models.

Which version is evaluated?

Each row names the latest open-weight version with a current LLMBoard score derived from eligible benchmark evidence.