ModelsEvery leading model, run where you choose
Pick the right model for every task
Choose from Context, Anthropic, OpenAI, Gemini, Kimi, DeepSeek and Qwen models in one switcher, add your own, and route every call through the gateway you trust.
Build the Q3 review deck from the updates doc and the valuation bridge. Same structure as Q2, the three flagged companies up front.
Claude Opus 5Most capableClaude Sonnet 5BalancedClaude Haiku 4.5Balanced · 200K context
GPT-5.6 SolGPT-5.4 mini
Gemini 3.1 Pro PreviewGemini 3 FlashFast
Kimi K3Kimi K2.6Balanced · 262K contextBuild the Q3 review deck from the updates doc and the valuation bridge. Same structure as Q2, the three flagged companies up front.
Claude Opus 5Most capableClaude Sonnet 5BalancedClaude Haiku 4.5Balanced · 200K contextGPT-5.6 SolGPT-5.4 miniGemini 3.1 Pro PreviewGemini 3 FlashFastUse any leading model
Every model your deployment offers sits in one switcher, grouped by maker, beside the task you are writing.
Build the Q3 review deck from the updates doc and the valuation bridge. Same structure as Q2, the three flagged companies up front.
Claude Opus 5Most capableClaude Sonnet 5BalancedClaude Haiku 4.5Balanced · 200K contextGPT-5.6 SolGPT-5.4 miniGemini 3.1 Pro PreviewGemini 3 FlashFastKimi K3Kimi K2.6Balanced · 262K contextAdd a model of your own
An admin adds a model with its provider, model id and credentials, tests it, and it is ready without a redeploy.
Models
Choose the defaults new agent sessions use and manage app-owned custom providers and model presentation. Running sessions pick changes up on their next launch.
- QQwen 3.8 Max (in-house)OpenAI-compatible · custom-llm/qwen-3.8-max · https://llm.internal/v1412msTestDelete
- Opus 5 (direct)Anthropic-compatible · custom-direct/opus-5 · https://api.anthropic.com688msTestDelete
Run it where you choose
Run a task in a managed sandbox or on a machine you paired, such as your own Mac, Windows or Linux computer.
Build the Q3 review deck from the updates doc and the valuation bridge. Same structure as Q2, the three flagged companies up front.
Choose the model for each task
The model switcher sits in the composer, so the choice is made where the work is described.
Build the Q3 review deck from the updates doc and the valuation bridge. Same structure as Q2, the three flagged companies up front.
Claude Opus 5Most capableClaude Sonnet 5BalancedClaude Haiku 4.5Balanced · 200K contextGPT-5.6 SolGPT-5.4 miniGemini 3.1 Pro PreviewGemini 3 FlashFastKimi K3Kimi K2.6Balanced · 262K contextBuild the Q3 review deck from the updates doc and the valuation bridge. Same structure as Q2, the three flagged companies up front.
Claude Opus 5Most capableClaude Sonnet 5BalancedClaude Haiku 4.5Balanced · 200K contextGPT-5.6 SolGPT-5.4 miniGemini 3.1 Pro PreviewGemini 3 FlashFastSet the models your company uses
Admins decide which models people see first, and which ones the company adds of its own.
Models
Choose the defaults new agent sessions use and manage app-owned custom providers and model presentation. Running sessions pick changes up on their next launch.
Gemini 3 FlashModels
Choose the defaults new agent sessions use and manage app-owned custom providers and model presentation. Running sessions pick changes up on their next launch.
Gemini 3 FlashRun agents where the work belongs
A model decides how an agent thinks. The machine decides where its work runs.
Build the Q3 review deck from the updates doc and the valuation bridge. Same structure as Q2, the three flagged companies up front.
Build the Q3 review deck from the updates doc and the valuation bridge. Same structure as Q2, the three flagged companies up front.
Secure by default
Every model call goes through the gateway your company chose, under the data rules you set.
Models
Choose the defaults new agent sessions use and manage app-owned custom providers and model presentation. Running sessions pick changes up on their next launch.
- QQwen 3.8 Max (in-house)OpenAI-compatible · custom-llm/qwen-3.8-max · https://llm.internal/v1412msTestDelete
- Opus 5 (direct)Anthropic-compatible · custom-direct/opus-5 · https://api.anthropic.com688msTestDelete
Models
Choose the defaults new agent sessions use and manage app-owned custom providers and model presentation. Running sessions pick changes up on their next launch.
- QQwen 3.8 Max (in-house)OpenAI-compatible · custom-llm/qwen-3.8-max · https://llm.internal/v1412ms
- Opus 5 (direct)Anthropic-compatible · custom-direct/opus-5 · https://api.anthropic.com688ms
Use the same models everywhere
The models your admins set up are the same in the web app, the phone apps, the desktop app and the terminal.
Build the Q3 review deck from the updates doc and the valuation bridge. Same structure as Q2, the three flagged companies up front.
Claude Opus 5Most capableClaude Sonnet 5BalancedClaude Haiku 4.5Balanced · 200K contextGPT-5.6 SolGPT-5.4 miniGemini 3.1 Pro PreviewGemini 3 FlashFastKimi K3Kimi K2.6Balanced · 262K contextBuild the Q3 review deck from the updates doc and the valuation bridge. Same structure as Q2, the three flagged companies up front.
Claude Opus 5Most capableClaude Sonnet 5BalancedClaude Haiku 4.5Balanced · 200K contextGPT-5.6 SolGPT-5.4 miniGemini 3.1 Pro PreviewGemini 3 FlashFastWork across the Context platform
Every product that runs an agent uses the models and gateways set up here.
- EngineThe agent runtime that plans the work, calls the model and uses the tools.
- EvalsScore agents on your own tasks before you change the model they use.
- Usage and billingSee spend by workspace, person, agent and model.
- Secure VMThe isolated sandbox each task runs in.
Home
- q3-portfolio-review.pptx1h ago
- q3-valuation-bridge.xlsx2h ago
- q3-portfolio-updates.docx3h ago
- lp-letter.pptx5h ago
Home
- q3-portfolio-review.pptx1h ago
- q3-valuation-bridge.xlsx2h ago
- q3-portfolio-updates.docx3h ago
- lp-letter.pptx5h ago
Deploy in your environment
Run Context in your own cloud and reach models through the gateway and private endpoints you already use.
- Your clusterContext runs on your own Kubernetes cluster, on Amazon EKS or Azure AKS.
- Your model gatewayReach models through the Context Gateway, LiteLLM, Amazon Bedrock, Azure OpenAI or a PrivateLink endpoint.
- Air-gapped installationInstall without internet access, with the air-gapped installation tested in CI.
- Your identity providerSign people in with SAML or OIDC and provision them with SCIM.
Context runs in production at Qualcomm. Read the case study
- Amazon EKS
- Azure AKS
- Air-gapped
Every Context workspace can use the models its deployment offers, including the Free plan.
- Free$0Includes Models
- Plus$20Per person per month
- EnterpriseCustomEnterprise deployments run in your own cloud.Talk to us about a plan
Questions
Which models can we use?
Context, Anthropic, OpenAI, Gemini, Kimi, DeepSeek and Qwen models, as offered by your deployment, plus any model your admins add.
Can we add our own model?
Yes. An admin adds it with its provider, model id and credentials, runs a test call, and it is available without a redeploy.
Can I change the model in the middle of a task?
Yes. Pick another model from the switcher, and the next turn uses it with everything the task has done so far.
How do we keep our data out of model training?
Require zero data retention. A model with no route that meets the requirement shows as unavailable and does not run.
Which gateways can we use?
The Context Gateway, LiteLLM, Amazon Bedrock, Azure OpenAI or a PrivateLink endpoint.
Where do agents run?
In a managed sandbox by default, or on a Mac, Windows or Linux machine you paired with Context.
