Model

Falcon-H1R 7B

OPEN

TII · Falcon-H1 family · released Jan 5, 2026

Reasoning-tuned hybrid model with DeepConf test-time scaling; matches models 2-7x its size.

ReasoningCodingVisionFunction callingTool useAgentic

1551.5

Elo · rank #110

Parameters

7.6B

Active params

Undisclosed

Context

256K tokens

Architecture

Hybrid Transformer + Mamba-2, reasoning-tuned (SFT on long CoT + GRPO RL, DeepConf test-time scaling)

License

TII Falcon-LLM License 2.0

Languages

—

API price (in/out)

No hosted API

Modalities

text

Benchmark results

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

AIMEMath88.1%#52

best: GPT-5.2 · 100.0%

GPQA DiamondReasoning61.3%#161

best: GPT-5.6 · 94.6%

Humanity's Last ExamReasoning11.1%#83

best: Claude Sonnet 5 · 57.4%

IFBenchReasoning53.4%#33

best: MiniMax M3 · 83.0%

LiveCodeBenchCoding68.6%#73

best: DeepSeek-V4-Pro (Think Max) · 93.5%

MATH-500Math97.4%#14

best: GPT-5 · 99.4%

MMLU-ProKnowledge72.1%#86

best: Claude Fable 5 · 91.5%

τ²-Bench TelecomAgents25.4%#37

best: Claude Opus 4.6 · 99.3%

Run it locally

VRAM @ Q4

4.7 GB

VRAM @ FP16

15.2 GB

Fits on (Q4)

RTX 3060 12GBRTX 4070 Ti 16GBRTX 3090 24GBRTX 4090 24GBRTX 5090 32GBM4 Pro 48GBM3 Max 128GBM3 Ultra 512GBA100 80GBH100 80GBH200 141GBB200 192GB

Vendor reports ~1,500 tok/s per GPU at batch size 64 in a server (vLLM) setting, not a single-user RTX 4090 llama.cpp measurement.

Quantizations

GGUF

Falcon-H1 family

Elo progression across releases

Falcon-H1 34BMay 2025939.0 Falcon-H1R 7BJan 20261551.5

Resources

HFtiiuae/Falcon-H1R-7B GHtiiuae/Falcon-H1 DOCSfalconllm.tii.ae/falcon-h1r-7b

API price weights · each benchmark row carries its own source badge (see methodology)