Model

Inkling

OPEN

Thinking Machines Lab · Inkling family · released Jul 15, 2026

Thinking Machines Lab's (Mira Murati's startup) first public model release — a 975B/41B-active MoE with adjustable inference-time "thinking effort", calibrated uncertainty, and text/vision/audio input.

ReasoningCodingVisionFunction callingTool useAgentic

2254.0

Elo · rank #38

Parameters

975B

Active params

41B (MoE)

Context

1M tokens

Architecture

Sparse MoE, 66 layers (975B total / 41B active, 6-of-256 routed experts + 2 shared experts/token), hybrid sliding-window/global attention (5:1), relative positional features (no RoPE)

License

Apache 2.0

Languages

—

API price (in/out)

$1.87 / $4.68

Modalities

text · vision · audio

Benchmark results

Bar shows position within the tracked field; marker = field best

AIMEMath97.1%#7

best: GPT-5.2 · 100.0%

BrowseCompAgents77.1%#18

best: Kimi K3 · 91.2%

CharXivVision78.1%#14

best: Muse Spark 1.1 (xhigh) · 88.4%

GDPval-AAAgents1238#26

best: Claude Fable 5 · 1932

GPQA DiamondReasoning87.2%#43

best: GPT-5.6 · 94.6%

Humanity's Last ExamReasoning29.7%#46

best: Claude Sonnet 5 · 57.4%

IFBenchReasoning79.8%#6

best: MiniMax M3 · 83.0%

MCP AtlasAgents74.1%#4

best: Kimi K3 · 84.2%

MMMU-ProVision73.5%#29

best: Claude Opus 4.7 · 85.5%

SimpleQAKnowledge43.9%#7

best: GPT-4.5 · 62.5%

SWE-bench ProCoding54.3%#28

best: Claude Fable 5 · 80.0%

SWE-bench VerifiedCoding77.6%#29

best: Claude Fable 5 · 95.0%

Terminal-Bench 2.0Coding63.8%#26

best: GPT-5.6 · 88.8%

Run it locally

VRAM @ Q4

600 GB

VRAM @ FP16

1900 GB

Fits on (Q4)

Multi-node cluster required

Official NVFP4 checkpoint needs Blackwell-class GPUs (~600GB VRAM); community Unsloth GGUF quants range 1-bit (~270GB) to 8-bit (~870GB) — far beyond a single RTX 4090 at any quant level.

Quantizations

NVFP4 · GGUF

Inkling family

Elo progression across releases

InklingJul 20262254.0

Resources

HFthinkingmachines/Inkling DOCShttps://thinkingmachines.ai/model-card/inkling/APIAvailable via Thinking Machines Lab API

API price $1.87/$4.68 · each benchmark row carries its own source badge (see methodology)