Google Gemini API model launch, deprecations and pricing changes, both batch and standard models

As of 12 August 2026, Google's Gemini 3.5 Pro model remains unreleased and in testing, missing the launch window suggested by recent industry reports. While the flagship model is delayed, Google announced that the Gemini ecosystem has surpassed one billion monthly users. Additionally, new deprecation dates have been confirmed for specialized models, and standard pricing for the Gemini 3.x family remains stable with significant discounts for batch processing.

• Gemini 3.5 Pro Release Status: Despite industry anticipation of a 12 August 2026 launch, official reports confirm that Gemini 3.5 Pro remains in testing and has not yet been released to the public or standard API tiers.
• Gemini User Milestone: On 12 August 2026, Google CEO Sundar Pichai announced that Gemini has reached one billion monthly users, making it the company's fastest-growing product to hit this milestone.
• Model Deprecation Schedule: The gemini-3.1-flash-image model is scheduled for deprecation on 17 August 2026, and the gemini-robotics-er-1.6-preview model will be shut down on 31 August 2026. 📅 Calendar file (.ics) Deprecation of gemini-3.1-flash-image August 17, 2026
• API Pricing and Batch Discounts: Standard pricing for Gemini 3.6 Flash is confirmed at $1.50 per 1M input tokens and $7.50 per 1M output tokens. The Batch API continues to offer a 50% cost reduction, though 'thinking tokens' used by reasoning models are billed at full output rates.

On 4 August 2026, Google announced a new Unified API for AI model routing, enabling developers to transcode and route traffic across different backend models through a single gateway. Additionally, Gemini Enterprise was updated with end-to-end tracing for data connectors to improve observability in agentic workflows. While Gemini 3.5 Pro remains in late-stage testing, official documentation continues to list it as 'coming soon' with industry reports pointing to a 12 August 2026 launch.

• Unified API for Model Routing: Google launched a public preview of a unified API gateway on 4 August 2026, allowing developers to send standard requests that are automatically transcoded to native Gemini or third-party backend schemas.
• Gemini Enterprise Tracing Support: As of 4 August 2026, Gemini Enterprise supports end-to-end tracing for data connectors, introducing new trace spans for tool execution and connector logic to visualize assistant-to-API workflows.
• Gemini 3.5 Pro Release Rumors: Reports on 4 August 2026 indicate that Gemini 3.5 Pro is expected to launch on 12 August 2026, featuring a 2-million-token context window and an enhanced 'Deep Think' reasoning mode. 📅 Calendar file (.ics) Expected Gemini 3.5 Pro Launch August 12, 2026
• Batch API Pricing: Google confirmed that the Batch API for Gemini 3.x models (including 3.6 Flash and 3.5 Flash-Lite) provides a 50% cost reduction on input and output tokens compared to standard tier pricing. (August 5, 2026) 📅 Calendar file (.ics) Gemini 3.x Batch Pricing Availability August 5, 2026
• Upcoming Image Model Shutdown: The gemini-3.1-flash-image and imagen-4.0-fast-generate-001 models are officially scheduled for shutdown on 17 August 2026. 📅 Calendar file (.ics) Gemini Image Model Shutdown August 17, 2026

Between 1 August and 3 August 2026, the unreleased Gemini 3.5 Pro model was spotted in the LMSYS Chatbot Arena for blind testing, indicating that Google has entered late-stage A/B evaluation for the flagship model. Additionally, Google confirmed specific deprecation deadlines for the month, including the shutdown of image generation and robotics preview models on 17 August and 31 August 2026, respectively.

• Gemini 3.5 Pro Arena Sightings: On 2 August 2026, developers reported that Gemini 3.5 Pro had appeared in the LMSYS Chatbot Arena's testing pool, suggesting an imminent public release despite official statements that the model remains in partner testing.
• Image Model Deprecation: Google has scheduled the shutdown of the gemini-3.1-flash-image and imagen-4.0-fast-generate-001 models for 17 August 2026. 📅 Calendar file (.ics) Gemini 3.1 Flash-Image and Imagen 4.0 Shutdown August 17, 2026
• Robotics Model Shutdown: The gemini-robotics-er-1.6-preview model is officially set to be shut down on 31 August 2026, following the transition to the ER 2 model family. 📅 Calendar file (.ics) Gemini Robotics ER 1.6 Shutdown August 31, 2026
• Sampling Parameter Deprecation: As of 1 August 2026, Google has officially deprecated the temperature, top_p, and top_k sampling parameters for the Gemini 3.x model family in favor of new reasoning controls.

In July 2026, Google significantly updated its API lineup by moving Gemini 3.6 Flash and Gemini 3.5 Flash-Lite to general availability, introducing a specialized cybersecurity model, and launching the Gemini Robotics ER 2 in public preview. While Gemini 3.6 Flash offers a 17% reduction in output token usage and lower pricing, the flagship Gemini 3.5 Pro remains in partner testing after missing its mid-year release target. Google also officially confirmed that pre-training has begun for its next-generation Gemini 4 models.

• Gemini 3.6 Flash and 3.5 Flash-Lite GA: On 21 July 2026, Google released stable, production-ready versions of Gemini 3.6 Flash and Gemini 3.5 Flash-Lite, with the 3.6 Flash model optimized for lower latency and a 17% reduction in output token usage.
• Gemini Robotics ER 2 Public Preview: Released on 30 July 2026, the Gemini Robotics ER 2 and its streaming variant provide advanced spatial reasoning and multi-robot coordination for embodied AI applications.
• Gemini 4 Pre-training Confirmation: Google officially confirmed on 21 July 2026 that it has commenced its 'most ambitious pre-training run yet' for the next-generation Gemini 4 model family.
• Gemini 3.6 Flash Pricing: Standard pricing for Gemini 3.6 Flash is set at $1.50 per 1M input tokens and $7.50 per 1M output tokens; the Batch API offers these at a 50% discount ($0.75/$3.75).
• Gemini API Parameter Deprecation: As of 21 July 2026, the sampling parameters temperature, top_p, and top_k are deprecated for the latest Gemini 3.x models in favor of newer optimization methods.
• Model Retirement Schedule: The Gemini 2.5 model family (Pro, Flash, and Flash-Lite) is scheduled for official retirement on 16 October 2026, while the gemini-robotics-er-1.6-preview will be shut down on 31 August 2026. 📅 Calendar file (.ics) Gemini 2.5 Family Retirement October 16, 2026