All platforms

One place for enterprise AI: routing, the gateway and Singularity Cloud.

Your teams are already vibe coding.

In every industry, people hand work to agents and chat models the company never approved, and the company’s data goes with it. Most companies don’t know. Singularity puts every model, team and request in one place you can see.

See how your teams work.

Every request through Singularity shows how a team uses AI: which work, which models, how often. Over time that becomes a map of the workflows worth automating, when the company chooses to.

  1. Observe

    Every model, team and request, in one view.

  2. Understand

    The workflows people repeat, and where they slow down.

  3. Automate, if you choose

    Turn a proven workflow into a system that runs.

Safe intelligence at every request.

The only way out to outside models, with every protection on it.

Checked before any model sees it.

The first failed check stops the request and says why.

  1. Identity and model accessWaitingPassed
  2. Application accessWaitingPassed
  3. Monthly capacityWaitingPassed
  4. Cost limitWaitingPassed
  5. Request frequencyWaitingPassed
  6. Knowledge permissionsWaitingPassed
  7. Source readinessWaitingPassed
  8. Model providerWaitingRan
Allowed and recordedSent to the chosen model, with its policy recorded.

Specialized agentic harnesses.

Custom agents for your industry and your company. Each runs in a Singularity harness around a frontier model, with the instructions, tools, rules and checks the work needs.

Frontier modelsAny model in the catalog, used as it is.
Singularity harness, for one industry
  • Instructions for the work
  • Tools and systems
  • The company’s rules and records
  • Checks on every output
Custom agents, built for that industry and that company.
Singularity gatewayIdentity, data protection, routing and spend, for every request.
Three layers: frontier models on top, the Singularity harness for one industry in the middle, and the Singularity gateway underneath.
The questionWhich suppliers need attention this quarter?
Frontier model aloneUnderstands the requestLists each supplier’s on-time rate.
Specialized harnessYour industry, your rules and recordsFlags suppliers under the company’s own on-time bar, with the record behind each.Company supplier criteria

Every frontier model.

Text, vision, audio, video and decisions, through one door.

  • TypeSafe
  • NVIDIA
  • OpenAI
  • Anthropic
  • Google
  • xAI
  • Meta
  • Mistral
  • DeepSeek
  • Qwen
  • Moonshot AI
  • Z.ai
  • MiniMax
  • ByteDance Seed
  • Xiaomi
  • Thinking Machines
  • Cohere
  • Amazon
  • Perplexity
  • Microsoft
  • IBM
  • Inception
  • Upstage
  • Sakana AI
  • Tencent
  • Poolside
  • View all models

One workspace, three ways to work.

Ask, build or hand work to an agent, without widening access.

Singularity
Supplier reviewWhich suppliers missed their delivery terms this quarter?Several approved suppliers missed terms, each cited to its register row and contract clause.Supplier registerReadyContract libraryReadySave to projectOpen in Code

Singularity Cloud

The gateway, hosted and run for you on leading cloud infrastructure.

  • Hosted for you

    On leading cloud infrastructure, run by our team.

  • Every model, routed

    One door to every model, for every team and request.

  • One place for all of it

    Routing, the gateway and its protections, together.

One gateway for every team.

Running first inside Keel Merchant Bank. Talk to us to bring it to your teams.

Start a conversation

Models

Provider

92 models

  • Jev Router

    typesafe/jev-router

    Picks a model and a reasoning effort for each request, on top of Jev.

    Modality
    Text, Image, Video, Audio, Files→Text
    Context
    1M
  • Jev Latest

    ~typesafe/jev-latest

    Always the newest Jev, TypeSafe’s System One model.

    Modality
    Text→Decisions
    Context
    32K
  • Jev 1.13

    typesafe/jev-1.13

    A System One model: it reads a situation and returns a decision.

    Modality
    Text→Decisions
    Context
    32K
  • Nemotron 3.5 Lightning

    nvidia/nemotron-3.5-lightning

    A small, fast open mixture of experts for high-volume agent work.

    Modality
    Text→Text
    Context
    262K
  • Nemotron 3.5 Content Safety

    nvidia/nemotron-3.5-content-safety

    A compact guardrail model that screens what goes into and comes out of other models.

    Modality
    Text, Image→Text
    Context
    131K
  • Nemotron 3 Ultra

    nvidia/nemotron-3-ultra-550b-a55b

    NVIDIA’s largest open model, for hard reasoning and for directing other agents.

    Modality
    Text→Text
    Context
    262K
  • Nemotron 3 Super

    nvidia/nemotron-3-super-120b-a12b

    An open hybrid mixture of experts tuned for multi-agent systems.

    Modality
    Text→Text
    Context
    262K
  • Nemotron 3 Nano 30B A3B

    nvidia/nemotron-3-nano-30b-a3b

    A small open model for building specialised agents cheaply.

    Modality
    Text→Text
    Context
    262K
  • Nemotron 3 Nano Omni

    nvidia/nemotron-3-nano-omni-30b-a3b-reasoning

    A small open model that reads text, images, video and audio for other agents.

    Modality
    Text, Image, Video, Audio→Text
    Context
    256K
  • GPT-6.1 Sol Pro

    openai/gpt-6.1-sol-pro

    GPT-6.1 Sol with more reasoning per answer.

    Modality
    Text, Image, Files→Text
    Context
    1.05M
  • GPT-6.1 Sol

    openai/gpt-6.1-sol

    The upgrade to GPT-6 Sol, one step below the flagship.

    Modality
    Text, Image, Files→Text
    Context
    1.05M
  • GPT-6 Astra Pro

    openai/gpt-6-astra-pro

    OpenAI’s flagship with more reasoning per answer.

    Modality
    Text, Image, Files→Text
    Context
    1.05M
  • GPT-6 Astra

    openai/gpt-6-astra

    OpenAI’s flagship, for demanding end-to-end work.

    Modality
    Text, Image, Files→Text
    Context
    1.05M
  • GPT-6 Sol

    openai/gpt-6-sol

    The high-end GPT-6 model at a lower cost than Astra.

    Modality
    Text, Image, Files→Text
    Context
    1.05M
  • GPT-6 Luna Pro

    openai/gpt-6-luna-pro

    GPT-6 Luna with more reasoning per answer.

    Modality
    Text, Image, Files→Text
    Context
    1.05M
  • GPT-6 Luna

    openai/gpt-6-luna

    The fast, low-cost tier of GPT-6.

    Modality
    Text, Image, Files→Text
    Context
    1.05M
  • GPT-5.6 Terra

    openai/gpt-5.6-terra

    The balanced middle of the GPT-5.6 line.

    Modality
    Text, Image, Files→Text
    Context
    1.05M
  • GPT-5.4 Mini

    openai/gpt-5.4-mini

    A faster GPT-5.4 for high-throughput work.

    Modality
    Text, Image, Files→Text
    Context
    400K
  • GPT-5.4 Nano

    openai/gpt-5.4-nano

    The lightest GPT-5.4, for speed and volume.

    Modality
    Text, Image, Files→Text
    Context
    400K
  • GPT-5.4 Image 2

    openai/gpt-5.4-image-2

    GPT-5.4 with image generation built in.

    Modality
    Text, Image, Files→Text, Image
    Context
    272K
  • GPT Audio

    openai/gpt-audio

    OpenAI’s audio model: speech in, speech out.

    Modality
    Text, Audio→Text, Audio
    Context
    128K
  • GPT-5.3-Codex

    openai/gpt-5.3-codex

    OpenAI’s agentic coding model.

    Modality
    Text, Image, Files→Text
    Context
    400K
  • gpt-oss-120b

    openai/gpt-oss-120b

    OpenAI’s larger open-weight model, for reasoning and agents.

    Modality
    Text→Text
    Context
    131K
  • gpt-oss-20b

    openai/gpt-oss-20b

    OpenAI’s small open-weight model, Apache 2.0.

    Modality
    Text→Text
    Context
    131K
  • gpt-oss-safeguard-20b

    openai/gpt-oss-safeguard-20b

    An open safety model that reasons over a policy you write.

    Modality
    Text→Text
    Context
    131K
  • Claude Opus 5.5

    anthropic/claude-opus-5.5

    Anthropic’s flagship, for hard reasoning, code and long agent work.

    Modality
    Text, Image, Files→Text
    Context
    1M
  • Claude Sonnet 5.5

    anthropic/claude-sonnet-5.5

    The everyday Claude, a direct upgrade on Sonnet 5.

    Modality
    Text, Image, Files→Text
    Context
    1M
  • Claude Fable 5.1

    anthropic/claude-fable-5.1

    Built for long-running agent work, code and knowledge work.

    Modality
    Text, Image, Files→Text
    Context
    1M
  • Claude Opus 5

    anthropic/claude-opus-5

    The previous Opus flagship.

    Modality
    Text, Image, Files→Text
    Context
    1M
  • Claude Sonnet 5

    anthropic/claude-sonnet-5

    The previous Sonnet, strong on code and agents.

    Modality
    Text, Image, Files→Text
    Context
    1M
  • Claude Fable 5

    anthropic/claude-fable-5

    The first Fable, for autonomous knowledge work and coding.

    Modality
    Text, Image, Files→Text
    Context
    1M
  • Claude Opus 4.8

    anthropic/claude-opus-4.8

    The last of the Opus 4 line.

    Modality
    Text, Image, Files→Text
    Context
    1M
  • Claude Haiku 4.5

    anthropic/claude-haiku-4.5

    The fastest Claude, close to frontier at a fraction of the cost.

    Modality
    Text, Image, Files→Text
    Context
    200K
  • Gemini 3.8 Flash

    google/gemini-3.8-flash

    Google’s strongest Flash model, across code, agents and reasoning.

    Modality
    Text, Image, Video, Audio, Files→Text
    Context
    1.05M
  • Gemini 3.7 Flash

    google/gemini-3.7-flash

    A fast multimodal model for agents and multi-step reasoning.

    Modality
    Text, Image, Video, Audio, Files→Text
    Context
    1.05M
  • Gemini 3.5 Flash

    google/gemini-3.5-flash

    Near-Pro coding and reasoning at Flash speed.

    Modality
    Text, Image, Video, Audio, Files→Text
    Context
    1.05M
  • Gemini 3.5 Flash Lite

    google/gemini-3.5-flash-lite

    The lightest Gemini, for volume.

    Modality
    Text, Image, Video, Audio, Files→Text
    Context
    1.05M
  • Gemini 3.1 Pro Preview

    google/gemini-3.1-pro-preview

    Google’s frontier reasoning model, in preview.

    Modality
    Text, Image, Video, Audio, Files→Text
    Context
    1.05M
  • Nano Banana 2

    google/gemini-3.1-flash-image

    Nano Banana 2: fast image generation and editing.

    Modality
    Text, Image→Text, Image
    Context
    131K
  • Nano Banana Pro

    google/gemini-3-pro-image

    Nano Banana Pro: Google’s most capable image model.

    Modality
    Text, Image→Text, Image
    Context
    131K
  • Gemma 4 31B

    google/gemma-4-31b-it

    Google DeepMind’s open dense model, text and images in.

    Modality
    Text, Image, Video→Text
    Context
    262K
  • Gemma 4 26B A4B

    google/gemma-4-26b-a4b-it

    An open mixture of experts from Google DeepMind.

    Modality
    Text, Image, Video→Text
    Context
    262K
  • Grok 4.7

    x-ai/grok-4.7

    xAI’s flagship for code, agents and knowledge work.

    Modality
    Text, Image, Files→Text
    Context
    500K
  • Grok 4.6

    x-ai/grok-4.6

    The previous Grok flagship, strong on STEM.

    Modality
    Text, Image, Files→Text
    Context
    500K
  • Grok 4.20 Multi-Agent

    x-ai/grok-4.20-multi-agent

    A Grok built to work as a team of agents.

    Modality
    Text, Image, Files→Text
    Context
    2M
  • Grok Build 0.1

    x-ai/grok-build-0.1

    A fast Grok trained for agentic software work.

    Modality
    Text, Image, Files→Text
    Context
    256K
  • Muse Spark 1.3

    meta/muse-spark-1.3

    Meta’s multimodal reasoning model for long agent work.

    Modality
    Text, Image, Video, Files→Text
    Context
    1.05M
  • Muse Glimmer 30B

    meta/muse-glimmer-30b

    An open model distilled from Muse Spark, small enough for local hardware.

    Modality
    Text, Image→Text
    Context
    131K
  • Llama 4 Maverick

    meta-llama/llama-4-maverick

    Meta’s large open Llama 4, text and images in.

    Modality
    Text, Image→Text
    Context
    1.05M
  • Llama 4 Scout

    meta-llama/llama-4-scout

    The smaller open Llama 4, with a very long context.

    Modality
    Text, Image→Text
    Context
    1.31M
  • Llama Guard 4 12B

    meta-llama/llama-guard-4-12b

    An open classifier that flags unsafe prompts and replies.

    Modality
    Text, Image→Text
    Context
    164K
  • Mistral Large 3 2512

    mistralai/mistral-large-2512

    Mistral’s most capable model, open weights.

    Modality
    Text, Image, Files→Text
    Context
    262K
  • Mistral Medium 3.5

    mistralai/mistral-medium-3-5

    A dense mid-size Mistral for instruction following.

    Modality
    Text, Image, Files→Text
    Context
    262K
  • Mistral Small 4

    mistralai/mistral-small-2603

    Several Mistral models folded into one small system.

    Modality
    Text, Image→Text
    Context
    262K
  • Devstral 2 2512

    mistralai/devstral-2512

    Mistral’s open model for agentic coding.

    Modality
    Text, Files→Text
    Context
    262K
  • Ministral 3 14B 2512

    mistralai/ministral-14b-2512

    The largest Ministral, small enough to run close to the data.

    Modality
    Text, Image→Text
    Context
    262K
  • Codestral 2508

    mistralai/codestral-2508

    Mistral’s code-completion model.

    Modality
    Text, Files→Text
    Context
    256K
  • Voxtral Small 24B 2507

    mistralai/voxtral-small-24b-2507

    Mistral Small with audio understanding.

    Modality
    Text, Audio, Files→Text
    Context
    33K
  • DeepSeek V4.1 Flash

    deepseek/deepseek-v4.1-flash

    DeepSeek’s fast mixture of experts on a new encoder-decoder design.

    Modality
    Text, Image→Text
    Context
    1.05M
  • DeepSeek V4 Pro 0813

    deepseek/deepseek-v4-pro-0813

    DeepSeek’s large mixture of experts.

    Modality
    Text→Text
    Context
    1.05M
  • DeepSeek V4 Flash Vision Exp

    deepseek/deepseek-v4-flash-vision-exp

    An experimental DeepSeek V4 Flash that reads images.

    Modality
    Text, Image→Text
    Context
    1.05M
  • Qwen3.8 Max Prime

    qwen/qwen3.8-max-prime

    Qwen’s largest model, served for higher throughput.

    Modality
    Text, Image, Video→Text
    Context
    1M
  • Qwen3.8 Omni Flash

    qwen/qwen3.8-omni-flash

    A Qwen that listens and watches as well as reads.

    Modality
    Text, Image, Video, Audio→Text
    Context
    1M
  • Qwen3.8 Flash

    qwen/qwen3.8-flash

    A fast multimodal Qwen for reasoning.

    Modality
    Text, Image, Video→Text
    Context
    1M
  • Qwen3.8 27B

    qwen/qwen3.8-27b

    An open dense Qwen for text, images and video.

    Modality
    Text, Image, Video→Text
    Context
    1M
  • Qwen3 Coder Next

    qwen/qwen3-coder-next

    An open Qwen for coding agents and local development.

    Modality
    Text→Text
    Context
    262K
  • Qwen3 VL 235B A22B Instruct

    qwen/qwen3-vl-235b-a22b-instruct

    An open Qwen that reads images and video closely.

    Modality
    Text, Image→Text
    Context
    262K
  • Kimi K3

    moonshotai/kimi-k3

    Moonshot’s open multimodal reasoning model.

    Modality
    Text, Image, Video→Text
    Context
    1.05M
  • Kimi K2.7 Code

    moonshotai/kimi-k2.7-code

    A Kimi for end-to-end programming over long contexts.

    Modality
    Text, Image→Text
    Context
    262K
  • GLM 5.3 Prime

    z-ai/glm-5.3-prime

    GLM-5.3 served faster.

    Modality
    Text→Text
    Context
    1M
  • GLM 5.3

    z-ai/glm-5.3

    Z.ai’s reasoning model for software and long agent tasks.

    Modality
    Text→Text
    Context
    1.05M
  • GLM 5V Turbo

    z-ai/glm-5v-turbo

    Z.ai’s multimodal model for vision-based coding and agents.

    Modality
    Text, Image, Video→Text
    Context
    203K
  • MiniMax M3

    minimax/minimax-m3

    MiniMax’s multimodal foundation model.

    Modality
    Text, Image, Video→Text
    Context
    1.05M
  • Seed 2.1 Turbo

    bytedance-seed/seed-2-1-turbo

    ByteDance’s multimodal model for code and agents.

    Modality
    Text, Image, Video→Text
    Context
    262K
  • Seed-2.0-Code

    bytedance-seed/seed-2.0-code

    ByteDance’s model for agentic coding.

    Modality
    Text, Image, Video→Text
    Context
    262K
  • MiMo-V2.6-Pro

    xiaomi/mimo-v2.6-pro

    Xiaomi’s flagship, reads text, images, video and audio.

    Modality
    Text, Image, Video, Audio→Text
    Context
    1.05M
  • Inkling

    thinkingmachines/inkling

    Thinking Machines’ open multimodal mixture of experts.

    Modality
    Text, Image, Audio→Text
    Context
    524K
  • Inkling Small

    thinkingmachines/inkling-small

    The smaller open Inkling.

    Modality
    Text, Image, Audio→Text
    Context
    524K
  • Command A+

    cohere/command-a-plus

    Cohere’s flagship for enterprise agent work.

    Modality
    Text, Image→Text
    Context
    192K
  • Command A

    cohere/command-a

    Cohere’s open-weights model for agents, many languages and code.

    Modality
    Text→Text
    Context
    256K
  • Nova 2 Lite

    amazon/nova-2-lite-v1

    A fast, low-cost Amazon model for everyday reasoning.

    Modality
    Text, Image, Video, Files→Text
    Context
    1M
  • Nova Premier 1.0

    amazon/nova-premier-v1

    Amazon’s most capable model, and a teacher for distilling smaller ones.

    Modality
    Text, Image→Text
    Context
    1M
  • Sonar Pro

    perplexity/sonar-pro

    Answers grounded in a live web search.

    Modality
    Text, Image→Text
    Context
    200K
  • Sonar Reasoning Pro

    perplexity/sonar-reasoning-pro

    Sonar with step-by-step reasoning over search results.

    Modality
    Text, Image→Text
    Context
    128K
  • Sonar Deep Research

    perplexity/sonar-deep-research

    Multi-step research across many sources.

    Modality
    Text→Text
    Context
    128K
  • Phi 4

    microsoft/phi-4

    A small Microsoft model that reasons well in little memory.

    Modality
    Text→Text
    Context
    16K
  • Granite 4.2 8B

    ibm-granite/granite-4.2-8b

    IBM’s small dense reasoning model.

    Modality
    Text→Text
    Context
    131K
  • Mercury 2.5

    inception/mercury-2.5

    A diffusion language model, built for speed.

    Modality
    Text→Text
    Context
    260K
  • Solar Pro 4

    upstage/solar-pro4

    Upstage’s cost-efficient model with a long context.

    Modality
    Text→Text
    Context
    524K
  • Fugu Ultra v2

    sakana/fugu-ultra-v2

    Sakana AI’s higher-performance Fugu.

    Modality
    Text, Image, Files→Text
    Context
    1M
  • Hy4 preview

    tencent/hy4-preview

    Tencent’s large mixture of experts, in preview.

    Modality
    Text→Text
    Context
    1.05M
  • Laguna S 2.1

    poolside/laguna-s-2.1

    Poolside’s coding agent model.

    Modality
    Text→Text
    Context
    1.05M

Facts from each provider’s public listing on OpenRouter. Model names and logos identify their providers.