Review · AISource: SpaceXSI
Grok Voice 3.0: migration details, pricing and reported benchmarks
Grok Voice 3.0 has automatically replaced Grok Voice 2.0 for applications using grok-voice-latest, according to the Speech to Speech API documentation. [1] The announcement is published under SpaceXSI branding; its footer names SpaceXAI LLC as copyright holder. [2] The provider says endpoints, session parameters and events remain unchanged. Developers configure voices and tools through session.update. [1]
Speech-to-speech API
Grok Voice 3.0
VerdictPromising
Provisional assessment based on provider documentation and reported benchmarks: Grok Voice 3.0 merits a pinned evaluation. Broader deployment depends on verified backend actions and billed cost per completed task.
- Best for
- Developers maintaining support or booking voice agents who can verify tool actions and compare pinned model versions
- Pricing
- $0.08 per minute ($4.80 per hour), plus $0.004 per text input [3]; a free trial is advertised, with allowance and eligibility unspecified [2]
Pros
- Unchanged endpoints, session parameters and events, according to the provider [1]
- Support for function tools, file search and web search [1]
- Same published unit pricing as Grok Voice 2.0 [3]
Cons
- Automatic model replacement for applications using grok-voice-latest [1]
- Uneven success across Cekura’s booking and Medicare scenarios [4]
- Unspecified retirement deadline for deprecated Grok Voice 2.0 [1]
Why it matters
For deployed agents, audit grok-voice-latest usage and pin the version under evaluation. Confirm that deprecated grok-voice-2.0 remains callable before using it as a baseline. [1] Cekura reports 97.7% booking-call success but 58.0% Medicare-call success, showing how results vary by workload. [4] SiliconANGLE reports that GPT-Live-1 leads Grok Voice 3.0 on the τ-Voice support benchmark, 74.5% to 71.6%. [6] These results do not establish lower cost per completed task, faster responses, or preserved interruption handling and tool reliability in a matched application test.
Attribution
Sources
- 1SpaceXSISpeech to Speech
- 2SpaceXSIIntroducing Grok Voice 3.0
- 3SpaceXSIAPI Pricing
- 4Cekura12 Best Speech-to-Speech (S2S) Models 2026 | Tested on Real Calls
- 5eesel AIGrok Voice 3.0 review: smarter on support calls, a beat slower
- 6SiliconANGLEGrok Voice 3.0 from SpaceXSI outscores OpenAI and Google on speech-to-speech index
Read the original source for full details and context.
Review · AI
Simplicity AI SDK beta connects coding agents to Silicon Labs BLE tools
Silicon Labs’ Simplicity AI SDK gives coding agents access to documentation, tools, and hardware for Bluetooth Low Energy (BLE) firmware development. The company announced the public beta on October 7, 2026. Its release notes label it Beta1 (September 2026). According to Silicon Labs, local and cloud Model Context Protocol (MCP) servers connect agents to development tools. Agents can inspect boards, modify projects, build, flash, and debug firmware. The SDK supports GitHub Copilot and Codex in VS Code, plus Cursor. Developers provide their own LLM through an editor or agent subscription.
Review · AI
Kokon previews local Qwen image editing with Windows and Apple Silicon installers
GabrieleLozupone’s Kokon repository describes local image editing with Qwen-Image 2.1 Turbo, automatic settings and installers for Windows and Apple Silicon Macs. Mac support is in beta. Its version 0.9.2 release claims native 2K editing on GPUs with at least 16 GB of video memory. It names stable-diffusion.cpp engine revision master-964-1d0a58a. The release also mentions an earlier version, 0.9.1.
AI
GitHub introduces HydraFusion Max and updates Copilot’s multi-model workflows
GitHub introduced HydraFusion Max for demanding tasks and changed how HydraFusion drafts and reviews responses. Both remain research previews in VS Code, the Copilot app and Copilot CLI.