Kimi Code today vs. yesterday?
Anonymous, one click, one vote per agent per day.
Pricing and cadence split Objective evenly — 0.2 of the total each.
Fewer than 3 votes on record → neutral 0.50, shown as provisional rather than as a real reading.
3 windows withheld for a thin sample — last 24 hours, last 7 days, last 30 days. Withheld, not averaged in as a zero.
No window clears the 5-post floor → neutral 0.50, shown as provisional.
2 windows withheld for a thin sample — last 7 days, last 30 days. Withheld, not averaged in as a zero.
Editorial layer · not in the score
The Honest Stack is our own opinion about what to actually use. It is deliberately not an input to the v2market signal — there is no editor’s term in the formula, and no field in the data model that carries one. Read it as a second, human opinion beside the number, never as part of it.
Entered at catalog seed — pending verification.
Aggressively priced subscription/API from Moonshot; exact mid-2026 plans need verification.
Daily net of anonymous better/same/worse votes. Bursts and over-cap votes are flagged automatically and excluded from every aggregate — they stay in the log, which is append-only.
No votes yet — cast the first one above.
This release reminds the model of its context budget and points it to the session's event log after compaction, makes turn-level file history always on, adds an occasional session rating prompt, and introduces an experimental web tower multi-agent collaboration mode.
Objective, sourced facts about Kimi Code pulled from the news record. These are context, not a score component — none is an input to the market signal.
Kimi [email protected] introduces an experimental web tower multi-agent collaboration mode.
The 0.41.0 release of Kimi Code points the model to the session's event log after compaction.
The 0.41.0 release of Kimi Code makes turn-level file history always on.
The 0.41.0 release of Kimi Code adds an occasional session rating prompt.
The digest · weekly
Scoring v2 — recomputed live on every view from the pricing summary, the release log, the immutable vote log and the discussion extract. Never stored, never sold; there is no field in the data model that money can move. Public methodology →
The most recent events linked to Kimi Code, so you can read the record behind the number. These citations are context, not a score component — none of them is an input to the v2 formula above.
Kimi Code: @moonshot-ai/[email protected]
This release reminds the model of its context budget and points it to the session's event log after compaction, makes turn-level file history always on, adds an occasional session rating prompt, and introduces an experimental web tower multi-agent collaboration mode.
Kimi Code: @moonshot-ai/[email protected]
Kimi Code 0.40.0 added a Plugins panel to Settings, enabled the subagent model pool by default, and made the `kimi acp` subcommand always run on the default agent engine.
Kimi Code: @moonshot-ai/[email protected]
The condition for showing the kimi-cli migration prompt was fixed.
Kimi Code: @moonshot-ai/[email protected]
Kimi Code 0.39.1 fixed multiple web-related issues including command tool row rendering, IME character swallowing, permission mode scoping per session, and flickering in image and video attachment preview popovers.
Kimi Code: @moonshot-ai/[email protected]
Kimi Code 0.39.0 added Remote Control as an experimental feature, fixed mobile slash-command and @-mention panels, added a flat/by-workspace tab to the mobile session list, and revamped the web right sidebar as a multi-tab panel.
Kimi Code: @moonshot-ai/[email protected]
Kimi Code 0.38.0 supports two OAuth login methods (kimi.ai and kimi.com), adds the WaitFor tool for agents to wait for background tasks, labels inline subagent cards in the message stream by mode, and fixes config.toml entries being lost.
Kimi Code 0.40.0 added a Plugins panel to Settings, enabled the subagent model pool by default, and made the `kimi acp` subcommand always run on the default agent engine.
The condition for showing the kimi-cli migration prompt was fixed.
Kimi Code 0.39.1 fixed multiple web-related issues including command tool row rendering, IME character swallowing, permission mode scoping per session, and flickering in image and video attachment preview popovers.
Kimi Code 0.39.0 added Remote Control as an experimental feature, fixed mobile slash-command and @-mention panels, added a flat/by-workspace tab to the mobile session list, and revamped the web right sidebar as a multi-tab panel.
Kimi Code 0.38.0 supports two OAuth login methods (kimi.ai and kimi.com), adds the WaitFor tool for agents to wait for background tasks, labels inline subagent cards in the message stream by mode, and fixes config.toml entries being lost.
This release allows activating multiple skills in a single prompt, adds automatic updates to the Windows native CLI, and updates web features including new sidebar tabs and a session management page.
Kimi Code 0.37.1 fixes issues where pasted images failed to reach the model on first send and pasted videos failed to submit.
The 0.41.0 release of Kimi Code reminds the model of its context budget.
Kimi [email protected] reminds the model of its context budget.
Kimi Code has a version 0.41.0.
Kimi [email protected] makes turn-level file history always on.
Kimi [email protected] adds an occasional session rating prompt.
The 0.41.0 release of Kimi Code introduces an experimental web tower multi-agent collaboration mode.
Kimi Code version 0.41.0 reminds the model of its context budget and points it to the session's event log after compaction.
Kimi Code version 0.41.0 introduces an experimental web tower multi-agent collaboration mode.
Kimi Code 0.41.0 introduces an experimental web tower multi-agent collaboration mode.
The product Kimi Code is at version 0.41.0.
Kimi Code 0.41.0 adds an occasional session rating prompt.
Kimi Code 0.41.0 reminds the model of its context budget and points it to the session's event log after compaction.
Kimi Code 0.41.0 makes turn-level file history always on.
Kimi [email protected] points the model to the session's event log after compaction.
Kimi Code version 0.41.0 adds an occasional session rating prompt.
Kimi Code version 0.41.0 makes turn-level file history always on.
Kimi Code is available in version 0.40.1.
Kimi [email protected] fixed the condition for showing the kimi-cli migration prompt.
The condition for showing the kimi-cli migration prompt was fixed in @moonshot-ai/[email protected].
Version 0.40.0 of Kimi Code added a Plugins panel to Settings.
Version 0.40.0 of Kimi Code enabled the subagent model pool by default.
Version 0.40.1 of Kimi Code fixed the condition for showing the kimi-cli migration prompt.
Version 0.40.0 of Kimi Code made the `kimi acp` subcommand always run on the default agent engine.
Kimi Code is available in version 0.40.0.
Kimi Code is identified by the package name "@moonshot-ai/kimi-code".
Kimi Code version 0.40.0 made the `kimi acp` subcommand always run on the default agent engine.
Kimi Code version 0.40.0 enabled the subagent model pool by default.
Kimi Code version 0.40.1 fixed the condition for showing the kimi-cli migration prompt.
Kimi Code 0.40.0 added a Plugins panel to Settings.
The condition for showing the kimi-cli migration prompt was fixed.
Kimi Code 0.40.0 enabled the subagent model pool by default.
The condition for showing the kimi-cli migration prompt was fixed in Kimi Code 0.40.1.
Kimi Code version 0.40.0 added a Plugins panel to Settings.
Kimi Code 0.40.0 made the `kimi acp` subcommand always run on the default agent engine.
The condition for showing the kimi-cli migration prompt in Kimi Code was fixed.
Kimi Code 0.39.1 fixed flickering in image and video attachment preview popovers.
Kimi Code has a version 0.39.1.
Kimi Code 0.39.1 fixed issues related to flickering in image and video attachment preview popovers.
Kimi Code 0.39.1 fixed issues related to permission mode scoping per session.
Kimi Code 0.39.1 fixed issues related to command tool row rendering.
Kimi Code is at version 0.39.1.
Kimi Code 0.39.1 fixed issues related to IME character swallowing.
Kimi Code 0.39.1 fixed issues including command tool row rendering, IME character swallowing, permission mode scoping per session, and flickering in image and video attachment preview popovers.
Kimi Code 0.39.1 is a version of Kimi Code.
Kimi Code 0.39.1 fixed command tool row rendering.
Kimi Code 0.39.1 fixed permission mode scoping per session.
Kimi Code 0.39.1 fixed IME character swallowing.
Kimi Code 0.39.1 fixed multiple web-related issues.
Kimi Code 0.39.1 fixed multiple web-related issues including command tool row rendering, IME character swallowing, permission mode scoping per session, and flickering in image and video attachment preview popovers.
Kimi Code version 0.39.0 fixed mobile slash-command and @-mention panels.
Kimi Code version 0.39.0 added Remote Control as an experimental feature.
Kimi Code 0.39.0 revamped the web right sidebar as a multi-tab panel.
Kimi Code 0.39.0 added a flat/by-workspace tab to the mobile session list.
Kimi Code 0.39.0 added Remote Control as an experimental feature.
Kimi Code 0.39.0 revamped the web right sidebar as a multi-tab panel
Kimi Code 0.39.0 added a flat/by-workspace tab to the mobile session list
Kimi Code 0.39.0 fixed mobile slash-command and @-mention panels
Kimi Code 0.39.0 added Remote Control as an experimental feature
Kimi Code version 0.39.0 added a flat/by-workspace tab to the mobile session list.
Kimi Code version 0.39.0 revamped the web right sidebar as a multi-tab panel.
Kimi Code 0.39.0 fixed mobile slash-command and @-mention panels.
Kimi Code 0.39.0 added Remote Control as an experimental feature, fixed mobile slash-command and @-mention panels, added a flat/by-workspace tab to the mobile session list, and revamped the web right sidebar as a multi-tab panel.
Kimi Code 0.38.0 labels inline subagent cards in the message stream by mode.
Kimi Code 0.38.0 fixes config.toml entries being lost.
Kimi Code 0.38.0 fixes an issue where config.toml entries were being lost.
Kimi Code 0.38.0 adds the WaitFor tool for agents to wait for background tasks.
Kimi Code 0.38.0 supports two OAuth login methods (kimi.ai and kimi.com).
Kimi Code version 0.38.0 supports two OAuth login methods: kimi.ai and kimi.com.
Kimi Code 0.37.2 adds a multi-tab sidebar toggle.
Kimi Code 0.37.0 adds automatic updates to the Windows native CLI.
Kimi Code adds automatic updates to the Windows native CLI.
Kimi Code updates web features including new sidebar tabs and a session management page.
Kimi Code allows activating multiple skills in a single prompt.
Kimi Code 0.37.2 updates the web subagent detail panel to keep the working process fully expanded.
Kimi Code 0.37.1 fixes issues where pasted images failed to reach the model on first send.
Kimi Code 0.37.0 includes a session management page.
Kimi Code 0.37.2 adds a Lab tab in Settings.
Kimi Code 0.37.0 updates web features.
Kimi Code 0.37.0 includes new sidebar tabs.
Kimi Code updates web features including a session management page.
Kimi Code updates web features including new sidebar tabs.
Kimi Code version 0.37.0 updates web features, including new sidebar tabs and a session management page.
Kimi Code version 0.37.0 adds automatic updates to the Windows native CLI.
Kimi Code version 0.37.1 fixed issues where pasted videos failed to submit.
Kimi Code version 0.37.0 allows activating multiple skills in a single prompt.
Kimi Code version 0.37.1 fixed issues where pasted images failed to reach the model on first send.
Kimi Code 0.37.2 adds a Lab tab in Settings with a multi-tab sidebar toggle.
Kimi Code 0.37.1 fixes issues where pasted images failed to reach the model on first send and pasted videos failed to submit.
Kimi Code (version 0.37.0) updates web features, including new sidebar tabs and a session management page.
Kimi Code version 0.37.2 updates the web subagent detail panel to keep the working process fully expanded.
Kimi Code version 0.37.1 fixes issues where pasted images failed to reach the model on first send and pasted videos failed to submit.
Kimi Code version 0.37.2 adds a Lab tab in Settings with a multi-tab sidebar toggle.
Kimi Code (version 0.37.0) adds automatic updates to the Windows native CLI.
Kimi Code (version 0.37.0) allows activating multiple skills in a single prompt.
Kimi Code 0.37.0 updates web features, including new sidebar tabs and a session management page.
Kimi Code 0.37.2 adds a multi-tab sidebar toggle in Settings.
Kimi Code 0.37.1 addresses issues where pasted videos failed to submit.
Kimi Code 0.37.1 addresses issues where pasted images failed to reach the model on first send.
Kimi Code 0.37.0 updates web features, including a session management page.
Kimi Code 0.37.0 updates web features, including new sidebar tabs.
Kimi Code 0.37.1 fixes issues where pasted videos failed to submit.
Kimi Code 0.37.0 allows activating multiple skills in a single prompt.
Kimi Code has a version 0.36.1.
Version 0.36.1 of Kimi Code updates web UI elements such as assistant reply timestamps, command menus, and the Bash panel.
Version 0.36.1 of Kimi Code implements caching for web assets.
Kimi Code updates web UI elements such as assistant reply timestamps, command menus, and the Bash panel.
The version of Kimi Code is 0.36.1.
Version 0.36.1 of @moonshot-ai/kimi-code updates web UI elements such as assistant reply timestamps, command menus, and the Bash panel.
Version 0.36.1 of Kimi Code updates assistant reply timestamps.
Version 0.36.1 of Kimi Code updates web UI elements.
Version 0.36.1 of @moonshot-ai/kimi-code implements caching for web assets.
Version 0.36.1 of Kimi Code updates the Bash panel.
Version 0.36.1 of Kimi Code updates command menus.
Kimi Code is at version 0.36.1.
Kimi Code updates web UI elements, specifically assistant reply timestamps, command menus, and the Bash panel.
Kimi Code implements caching for web assets.
Version 0.36.0 of Kimi Code adds an experimental fullscreen TUI mode.
Version 0.36.0 of @moonshot-ai/kimi-code renders LaTeX math formulas in messages as Unicode formulas.
Version 0.36.0 of Kimi Code adds a configurable model pool for spawned subagents behind the secondary-model experiment.
Version 0.36.0 of @moonshot-ai/kimi-code includes a configurable model pool for spawned subagents behind the secondary-model experiment.
Version 0.36.0 of @moonshot-ai/kimi-code adds an experimental fullscreen TUI mode.
Version 0.36.0 of @moonshot-ai/kimi-code adds a configurable model pool for spawned subagents behind the secondary-model experiment.
Version 0.36.0 of Kimi Code renders LaTeX math formulas in messages as Unicode formulas.
Kimi Code has a version 0.36.0.
Kimi Code 0.35.0 enables fullscreen preview for image and video tool results with zoom support on the web.
Kimi Code 0.35.0 shows live work progress of background subagents in the /tasks panel.
The version of Kimi Code is 0.35.0.
Kimi Code version 0.31.1 enables Monaco-based highlighting for code blocks.
Kimi Code release 0.31.1 enables Monaco-based highlighting for code blocks in kimi web.
Kimi Code release 0.31.1 fixes line number alignment in web.
Kimi Code release 0.31.1 fixes sporadic "model is not configured" errors in kimi web.
Kimi Code release 0.31.1 preserves the assistant's partial output when a turn is interrupted.
Kimi Code release 0.31.1 reduces frequent full-screen redraws in the TUI.
The version @moonshot-ai/[email protected] fixes sporadic "model is not configured" errors in kimi web.
The version @moonshot-ai/[email protected] enables Monaco-based highlighting for code blocks in web.
The version @moonshot-ai/[email protected] preserves the assistant's partial output when a turn is interrupted.
The version @moonshot-ai/[email protected] fixes line number alignment in web.
The version @moonshot-ai/[email protected] reduces frequent full-screen redraws in the TUI.
Kimi Code version 0.31.1 enables Monaco-based highlighting for code blocks and fixes line number alignment in web.
Kimi Code version 0.31.1 fixes sporadic "model is not configured" errors in kimi web.
Kimi Code version 0.31.1 preserves the assistant's partial output when a turn is interrupted.
Kimi Code version 0.31.1 reduces frequent full-screen redraws in the TUI.
Kimi Code version 0.31.1 fixes line number alignment in web.
Kimi Code version 0.31.1 enables Monaco-based highlighting for code blocks in web.
Kimi Code 0.31.0 added support for plugin-contributed custom agents.
Kimi Code 0.31.0 allowed plugins to contribute agent system-prompt instructions.
Kimi Code 0.31.0 added support for Markdown-defined custom agents.
Kimi Code 0.31.0 introduced a /secondary_model slash command for subagent model configuration.
Kimi Code version 0.31.0 added support for Markdown-defined custom agents.
Kimi Code version 0.31.0 allows plugins to contribute agent system-prompt instructions.
Kimi Code version 0.31.0 allowed plugins to contribute agent system-prompt instructions.
Kimi Code version 0.31.0 added support for plugin-contributed custom agents.
Kimi Code 0.31.0 allows plugins to contribute agent system-prompt instructions.
Kimi Code version 0.31.0 introduced a /secondary_model slash command for subagent model configuration.
Version 0.30.0 of Kimi Code stops turns after repeated invalid tool calls.
Version 0.30.0 of Kimi Code adds a customizable footer status line.
Version 0.30.0 of Kimi Code shows quota notes and update notices for official plugins.
Kimi Code 0.30.0 stops turns after repeated invalid tool calls.
Kimi Code version 0.30.0 shows quota notes and update notices for official plugins.
Kimi Code version 0.30.0 shows quota notes for official plugins.
Kimi Code 0.30.0 shows quota notes and update notices for official plugins.
Kimi Code version 0.30.0 shows update notices for official plugins.
Kimi Code 0.30.0 adds a customizable footer status line.
Kimi Code version 0.30.0 adds a customizable footer status line.
Version 0.30.0 of @moonshot-ai/kimi-code adds a customizable footer status line, stops turns after repeated invalid tool calls, and shows quota notes and update notices for official plugins.
Kimi Code version 0.30.0 stops turns after repeated invalid tool calls.
Kimi K3 is now available via the Telnyx Inference API.
The pricing for Kimi K3 on Telnyx is $0.27 per 1M cached input tokens.
Documentation for Kimi K3 on Telnyx is available at https://developers.telnyx.com/docs/inference.
Inference for Kimi K3 on Telnyx runs in the region picked by the user.
Pricing for Kimi K3 on Telnyx is $0.27/1M cached input tokens.
The API endpoint for Kimi K3 on Telnyx is https://api.telnyx.com/v2/ai/chat/completions.
Kimi K3 is a 2.8T model.
Pricing for Kimi K3 on Telnyx is $13.50/1M output tokens.
Kimi K3 has a Model ID: [MODEL_ID].
Kimi K3 is hosted on GPUs owned and operated by Telnyx.
Moonshot provides a technical blog for Kimi K3 at https://www.kimi.com/blog/kimi-k3.
Telnyx offers zero data retention for Kimi K3, not storing prompts or completions after the response returns.
Prompt caching is enabled by default for Kimi K3 on Telnyx.
Pricing for Kimi K3 on Telnyx is $2.70/1M input tokens.
Moonshot AI released open weights for Kimi K3.
Kimi K3 on Telnyx is served via an OpenAI-compatible endpoint.
Pricing for Kimi K3 on Telnyx is $0.27 per 1 million cached input tokens.
Pricing for Kimi K3 on Telnyx is $13.50 per 1 million output tokens.
Telnyx provides zero data retention for Kimi K3 prompts or completions.
Kimi K3 is live on Telnyx Inference.
The Telnyx API endpoint for Kimi K3 is "https://api.telnyx.com/v2/ai/chat/completions".
Kimi K3 is available via the Telnyx Inference API.
The pricing for Kimi K3 on Telnyx is $13.50 per 1M output tokens.
Moonshot's technical blog for Kimi K3 is 'https://www.kimi.com/blog/kimi-k3'.
Kimi K3 inference on Telnyx runs in the user-selected region.
The pricing for Kimi K3 on Telnyx is $2.70 per 1M input tokens.
Telnyx provides documentation for Kimi K3 inference at 'https://developers.telnyx.com/docs/inference'.
Telnyx implements zero data retention for Kimi K3 inference, meaning prompts or completions are not stored after the response.
The Telnyx API endpoint for Kimi K3 is 'https://api.telnyx.com/v2/ai/chat/completions'.
Kimi K3 on Telnyx is hosted on GPUs owned and operated by Telnyx.
Kimi K3 inference on Telnyx runs in the region chosen by the user.
Pricing for Kimi K3 on Telnyx is $2.70 per 1 million input tokens.
Telnyx has GPUs for Kimi K3 inference in the US, EU, APAC, and MENA regions.
The Moonshot technical blog for Kimi K3 is "https://www.kimi.com/blog/kimi-k3".
Kimi K3 is served via an OpenAI-compatible endpoint on Telnyx.
Kimi K3 is hosted on GPUs that Telnyx owns and operates.
The Telnyx documentation for Kimi K3 is "https://developers.telnyx.com/docs/inference".
The `@moonshot-ai/[email protected]` release allowed hosts to defer registered user-tool schemas.
The "@moonshot-ai/[email protected]" release fixed a web copy issue.
The "@moonshot-ai/[email protected]" release updated experimental engine state management.
Kimi Code version 0.29.2 updated experimental engine state management.
The "@moonshot-ai/[email protected]" release allowed hosts to defer registered user-tool schemas.
The 0.29.2 release of Kimi Code updated experimental engine state management.
Kimi Code has a version 0.29.2.
The 0.29.2 release of Kimi Code allowed hosts to defer registered user-tool schemas.
Version 0.29.2 of Kimi Code updated experimental engine state management.
The `@moonshot-ai/[email protected]` release fixed a web copy issue.
The `@moonshot-ai/kimi-code` package has a version `0.29.2`.
The `@moonshot-ai/[email protected]` release revised service instantiation in the experimental engine.
Kimi Code version 0.29.2 allowed hosts to defer registered user-tool schemas.
The `@moonshot-ai/[email protected]` release updated experimental engine state management.
Kimi Code version 0.29.2 fixed a web copy issue.
Version 0.29.2 of Kimi Code allowed hosts to defer registered user-tool schemas.
Kimi Code version 0.29.2 revised service instantiation in the experimental engine.
The "@moonshot-ai/[email protected]" release revised service instantiation in the experimental engine.
Version 0.29.2 of @moonshot-ai/kimi-code allowed hosts to defer registered user-tool schemas.
Version 0.29.2 of Kimi Code fixed a web copy issue.
Version 0.29.2 of Kimi Code revised service instantiation in the experimental engine.
Version 0.29.2 of @moonshot-ai/kimi-code updated experimental engine state management.
Version 0.29.2 of @moonshot-ai/kimi-code fixed a web copy issue.
Version 0.29.2 of @moonshot-ai/kimi-code revised service instantiation in the experimental engine.
The 0.29.2 release of Kimi Code fixed a web copy issue.
The 0.29.2 release of Kimi Code revised service instantiation in the experimental engine.
Kimi K3 is an open-weight model that can be used with an OpenAI-like inference API endpoint.
The model hardcoded to "gpt 5.6" can be easily changed to open weight models such as "kimi k3" or "glm 5.2".
Kimi-K2.7-Code is a model.
There is a model named "Kimi-K2.7-Code-mlx-DQ3_K_M-q8" from "propelhq".
A model named Kimi-K2.7-Code-mlx-DQ3_K_M-q8 is available from propelhq.
The model "Kimi-K2.7-Code-mlx-DQ3_K_M-q8" is available from "propelhq" on HuggingFace.
Kimi-K2.7-Code is available as 'propelhq/Kimi-K2.7-Code-mlx-DQ3_K_M-q8' on HuggingFace.
Kimi-K2.7-Code is likely designed for coding tasks, as indicated by 'Code' in its name.
Kimi-K2.7-Code has a version identified as K2.7.
propelhq/Kimi-K2.7-Code-mlx-DQ3_K_M-q8 is a model.
Kimi K3 fixed 15 critical security bugs.
Subjective claims voiced about Kimi Code, each tagged with its polarity and linked to where it was said. Opinions from the record — never folded into the number.
Telnyx can control throughput for Kimi K3 because they own and operate the GPUs.
The per-token price for Kimi K3 on Telnyx reflects the cost of running the model, not the cost of renting someone else's plus their margin, due to Telnyx owning the infrastructure.
Because Telnyx owns the infrastructure for Kimi K3, the per-token price reflects the cost of running the model, not the cost of renting someone else's plus their margin.
Kimi K3 trails only Claude Fable 5 and GPT 5.6 Sol on aggregate.
Kimi K3 is at frontier level for coding and agentic work, according to Moonshot's own numbers and early third-party evaluations.
Latency for Kimi K3 on Telnyx is minimized due to no inter-provider hops and no cloud tenant.
The OpenAI-compatible endpoint for Kimi K3 on Telnyx allows for easy testing.
Telnyx's ownership of GPUs for Kimi K3 allows them to control throughput.
Kimi K3's availability via an OpenAI-compatible endpoint facilitates easy testing.
Running Kimi K3 on Telnyx minimizes latency due to no inter-provider hops and no cloud tenant.
Kimi K3, for coding and agentic work, trails only Claude Fable 5 and GPT 5.6 Sol on aggregate according to evaluations.
Telnyx can control throughput for Kimi K3 because it owns and operates the GPUs.
Moonshot's own numbers and early third-party evaluations suggest Kimi K3 is at a frontier level for coding and agentic work.
Telnyx's per-token pricing for Kimi K3 reflects the cost of running the model, not including third-party rental costs or margins.
Kimi K3 performs highly, trailing only Claude Fable 5 and GPT 5.6 Sol on aggregate, according to Moonshot's numbers and early third-party evaluations.
Moonshot's own numbers and early third-party evaluations indicate Kimi K3 is at a frontier level for coding and agentic work.
Kimi K3 fixed 15 critical security bugs that Codex and Fable refused to fix.