Explore models.

2,813 models1,615 with prices34 metricsUpdated Oct 18, 2026 ↗

204 models · newest firstPage 1 / 9

More filters

deepseek-aiListed 1 day ago

DeepSeek-V4.1-Flash-1017

DeepSeek-V4.1-Flash-1017 is a minor update to DeepSeek-V4.1-Flash, delivering substantial improvements across reasoning, coding, agentic tasks, and visual benchmarks. Alongside these gains, the model…

1M contextImage inputTool calling

LiveBenchEffort: max · single run · equal mean of 7 suites79.2%

$0.05in$0.60out/ 1M tokens · USDRelace · via OpenRouter · 15 endpointsrelace

+Compare

xingchen-agiListed 1 day ago

XingChen: Xing4.0 29B A4B (free)

Apache-2.0 weights with switchable thinking; supports domain fine-tuning and vLLM, SGLang or KTransformers deployment.

262.1K context29BTool callingStructured outputNo reported score

No reported score

$0.00in$0.00out/ 1M tokens · USDPhala · via OpenRouter · 2 endpointsphala

+Compare

reflectionReleased 2 days ago

Reflection: Beam 1.0 501B A23B

Apache-2.0 text model with 501B total and 23B active parameters; ships BF16, FP8 and NVFP4 weights.

1M context501BTool callingStructured output

AA IntelligenceEffort not specified30.33 LiveBenchEffort: max · single run · equal mean of 7 suites55.5%

$0.25in$1.00out/ 1M tokens · USDNebius · via OpenRouter · 5 endpointsnebius/fp8 · quantization=fp8

+Compare

moondreamListed 2 days ago

Moondream: Moondream 3 Preview

Moondream 3 Preview is a vision-language model from Moondream with 9B total parameters and 2B active (mixture-of-experts). It handles visual question answering, captioning, OCR, and visual reasoning over…

32K contextImage inputReasoningNo reported score

No reported score

$0.30in$2.50out/ 1M tokens · USDPrimeIntellect · via OpenRouter · 2 endpointsprimeintellect/fp8 · quantization=fp8

+Compare

moondreamListed 2 days ago

Moondream: Moondream 3.1

Native object detection and pointing return structured coordinates; runs locally on Apple Silicon or NVIDIA GPUs.

32K context9BImage inputReasoningNo reported score

No reported score

$0.30in$2.50out/ 1M tokens · USDPrimeIntellect · via OpenRouter · 2 endpointsprimeintellect/fp8 · quantization=fp8

+Compare

metaReleased 3 days ago

Meta: Muse Spark Open

OpenRouter lists a 1,048,576-token context and up to 943,718 output tokens for this open-weight model aimed at agentic coding.

1M contextImage inputAudio input

AA IntelligenceEffort: xhigh · Xhigh38.87 LiveBenchEffort: max · single run · equal mean of 7 suites57.57%

$1.25in$4.25out/ 1M tokens · USDbaseten · via Hugging Face · 8 endpoints

+Compare

liquidListed 5 days ago

LiquidAI: LFM2.5-24B-A2B

24B total, 2.45B active parameters; always reasons before answering, with GGUF and MLX releases for local deployment.

128K context24BTool callingStructured outputNo reported score

No reported score

$0.05in$0.70out/ 1M tokens · USDLiquid · via OpenRouter · 2 endpointsliquid/fp8 · quantization=fp8

+Compare

inclusionAIListed 5 days ago

Ling-3.1-flash

Switch reasoning on or off; hybrid linear attention and multi-token prediction support long-context self-hosted inference.

262.1K contextOpen weightsTool callingNo reported score

No reported score

$0.30in$0.90out/ 1M tokens · USDDeepInfra · 2 endpoints

+Compare

nvidiaListed 6 days ago

NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-NVFP4

1M context120BImage inputOpen weightsNo reported score

No reported score

$0.20in$0.50out/ 1M tokens · USDbaseten · via Hugging Face

+Compare

xiaomiListed 27 days ago

Xiaomi: MiMo-V2.6-Flash

MiMo-V2.6-Flash is an open-source foundation model developed by Xiaomi. Built on a Mixture-of-Experts architecture with 309B total parameters and 15B activated per token, it employs a hybrid attention…

1.1M contextImage inputAudio inputNo reported score

No reported score

$0.10in$0.28out/ 1M tokens · USDDarkbloom · via OpenRouter · 10 endpointsdarkbloom/fp4 · quantization=fp4

+Compare

xiaomiReleased 27 days ago

Xiaomi: MiMo-V2.6-Pro

MiMo-V2.6-Pro is the flagship foundation model developed by Xiaomi. Built at a scale of over 1T parameters, it is designed to push the ceiling of capability for the most demanding...

1.1M contextImage inputAudio input

AA IntelligenceEffort not specified46.32 LiveBenchEffort: high · single run · equal mean of 7 suites74.87%

$0.43in$0.87out/ 1M tokens · USDDeepInfra · via OpenRouter · 6 endpointsdeepinfra/fp8 · quantization=fp8

+Compare

prism-mlListed 30 days ago

PrismML: Ternary Bonsai 2 27B

Bonsai 2 27B is a 27B-parameter reasoning model from PrismML derived from Qwen3.8-27B. It supports coding, mathematics, tool calling, and image understanding with a 262K-token context window. Ternary…

262.1K context27BImage inputTool callingNo reported score

No reported score

$0.075in$0.50out/ 1M tokens · USDDarkbloom · via OpenRouter · 2 endpointsdarkbloom/int4 · quantization=int4

+Compare

stepfun-aiReleased 30 days ago

Step-5-Preview

600B total and 27B active parameters; accepts images and video, with FP8 and BF16 weights for deployment.

1M contextImage inputVideo input

AA IntelligenceEffort not specified43.73 LiveBenchEffort: max · single run · equal mean of 7 suites62.01%

$1.00in$2.70out/ 1M tokens · USDDeepInfra · 6 endpoints

+Compare

nvidiaListed Sep 16

NVIDIA-Nemotron-3.5-Super-VL

NVIDIA Nemotron 3.5 Super VL is an open multimodal model for agents that see, reason, and act. Its 120B-parameter Mixture-of-Experts design (12B active) understands text, images, documents, GUIs, and video,…

1M contextImage inputVideo inputNo reported score

No reported score

$0.10in$0.40out/ 1M tokens · USDCrusoe · via OpenRouter · 7 endpointscrusoe/bf16 · quantization=bf16

+Compare

inference-netListed Sep 12

Inference.net: Schematron V2 Turbo

Schematron V2 Turbo is a 3B-parameter HTML-to-JSON extraction model from Inference.net. It prioritizes throughput for high-volume extraction workloads. Extraction instructions must be supplied through a…

128K contextStructured outputPrompt cachingNo reported score

No reported score

$0.03in$0.15out/ 1M tokens · USDInferenceNet · via OpenRouter · 2 endpointsinference-net/bf16 · quantization=bf16

+Compare

inference-netListed Sep 12

Inference.net: Schematron V2 Small

Schematron V2 Small is a 3B-parameter HTML-to-JSON extraction model from Inference.net. It prioritizes extraction quality for complex schemas and long pages. Extraction instructions must be supplied through…

128K contextStructured outputPrompt cachingNo reported score

No reported score

$0.05in$0.23out/ 1M tokens · USDInferenceNet · via OpenRouter · 2 endpointsinference-net/bf16 · quantization=bf16

+Compare

deepseekListed Sep 10

DeepSeek: DeepSeek V4.1 Flash (batch)

DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...

1M contextImage inputTool callingNo reported score

No reported score

$0.05in$0.15out/ 1M tokens · USDSail Research · via OpenRouter · 3 endpointssail-research/fp4 · quantization=fp4

+Compare

deepseekReleased Sep 10

DeepSeek: DeepSeek V4.1 Flash

Causal encoder-decoder activates 8B parameters for input and 16B for output, reducing the compute needed for long inputs.

1M contextImage inputTool calling

AA IntelligenceEffort: max · Max39.46 LiveBenchEffort: max · single run · equal mean of 7 suites78.04%

$0.04in$0.144out/ 1M tokens · USDOpenInference · via OpenRouter · 44 endpointsopen-inference/fp4 · quantization=fp4

+Compare

nex-agiListed Sep 8

Nex AGI: Nex-N2.5-Mini

Nex-N2.5 is an agentic model built to turn goals into working, verified outcomes. Its core strength is agentic coding within a visual feedback loop: it can explore codebases, implement multi-file...

262.1K contextImage inputStructured outputNo reported score

No reported score

$0.025in$0.10out/ 1M tokens · USDNex AGI · via OpenRouter · 2 endpointsnex-agi/bf16 · quantization=bf16

+Compare

nex-agiListed Sep 8

Nex AGI: Nex-N2.5-Pro

Nex-N2.5 is an agentic model built to turn goals into working, verified outcomes. Its core strength is agentic coding within a visual feedback loop: it can explore codebases, implement multi-file...

262.1K contextImage inputTool callingNo reported score

No reported score

$0.075in$0.25out/ 1M tokens · USDNex AGI · via OpenRouter · 2 endpointsnex-agi/fp8 · quantization=fp8

+Compare

inclusionAIListed Sep 4

Ling-3.0-flash-VL

The multimodal version built on Ling-3.0-flash — 124B total / ~5.5B active per token, with native text, image, and video understanding. It’s mainly designed for multimodal agentic workflows, long-context…

262.1K contextImage inputVideo inputNo reported score

No reported score

$0.06in$0.18out/ 1M tokens · USDdeepinfra · via Hugging Face · 5 endpoints

+Compare

inclusionAIListed Sep 3

Ling-3.0-flash-Fin

262.1K contextOpen weightsTool callingNo reported score

No reported score

$0.06in$0.18out/ 1M tokens · USDdeepinfra · via Hugging Face · 2 endpoints

+Compare

tencentListed Aug 27

Hy4-preview

Hy4 preview is a 770B-parameter Mixture-of-Experts flagship model from the Tencent Hy Team, with 49B parameters activated per token and a 1M-token context window. It uses Gated DeepSeek Sparse Attention…

1M contextTool callingStructured outputNo reported score

No reported score

$0.834in$2.501out/ 1M tokens · USDdeepinfra · via Hugging Face · 8 endpoints

+Compare

qwenListed Aug 26

Qwen: Qwen3.8 Flash

Qwen3.8 Flash is a multimodal reasoning model from Alibaba. It is suited for coding assistance, agentic workflows, visual understanding, document and codebase analysis, desktop interaction, chart analysis,…

1M contextImage inputVideo input

LiveBenchEffort: max · single run · equal mean of 7 suites68.79%

$0.15in$0.47out/ 1M tokens · USDAlibaba · via OpenRouter · 3 endpointsalibaba/us-east-1

+Compare