# Google is turning Managed Agents into scheduled workers with budgets and hooks Google has pushed its Gemini API Managed Agents closer to something builders can run as infrastructure rather than demos: Gemini 3.7 Flash is now the default, hooks can gate tool calls, token budgets can cap runaway work, and scheduled triggers can keep the same sandbox working over time. Gemini API Managed Agents now combine Gemini 3.7 Flash by default with environment hooks, token budgets, scheduled triggers and persistent sandboxes — a much more production-shaped agent runtime. - Status: Active - Published: 2026-08-13T22:13:36+12:00 - Updated: 2026-08-15T08:52:22+12:00 - Categories: Artificial Intelligence, Web Development - Tags: AI agents, Developer tools, Gemini, Gemini 3.7 Flash, Gemini API, Google, Managed Agents - Canonical HTML: https://beyondthe.news/dossiers/google-managed-agents-budgets-hooks-schedules ## Latest material change Google's current Antigravity Agent docs now make Gemini 3.7 Flash the default instead of 3.6 Flash. This proposal updates the runtime baseline, pricing context and builder guidance around explicitly pinning models; hooks, budgets and scheduled-worker conclusions remain intact. ## What changed Google has expanded Managed Agents in the Gemini API with several controls that make long-running agents easier to operate as real software. Environment hooks can gate tool calls, token budgets can cap an agent's total consumption, scheduled triggers can reuse persistent sandboxes, and the Environments API can inspect or delete those sandboxes. Since BTN's published version, Google has also changed the default model behind the Antigravity managed-agent runtime. Google's current Antigravity Agent documentation, last updated August 13, 2026, says `antigravity-preview-05-2026` now defaults to Gemini 3.7 Flash when `agent_config.model` is omitted; Gemini 3.6 Flash remains available as an explicit model choice. The earlier July 28 launch material had made 3.6 Flash the default. That default-model change matters operationally because builders who rely on the implicit default can receive a capability and behavior change without changing the agent identifier or their own code. Google currently prices Gemini 3.7 Flash and 3.6 Flash at the same promotional standard API rate through December 31, 2026, so the switch does not itself raise the published per-token rate today, but it can still change task success, latency and total token usage. ## Why it matters Agent products are increasingly becoming runtime products. A strong model is only one part of the system; developers also need a place where the agent can work, limits on what it can spend, controls over what it can do and a reliable way to run the same task again later. Google is bundling more of that operational layer into the Gemini API itself. A builder can ask a managed agent to work in an isolated environment, intercept file writes or code execution with custom checks, stop a run when a token budget is exhausted and schedule recurring work without first building a separate orchestration service. The switch from Gemini 3.6 Flash to Gemini 3.7 Flash also makes model pinning part of runtime governance. If repeatability matters, relying on the managed agent's implicit default means Google can move the underlying model as the platform evolves. Explicitly selecting `agent_config.model`, and re-running production evaluations before changing it, gives teams a more stable operating baseline. ## Hooks are the most useful addition Environment hooks can run custom commands or HTTP handlers before or after tool execution inside the managed sandbox. A pre-tool hook can deny a write or code-execution action, while a post-tool hook can lint, inspect or audit what the agent just produced. That matters because prompt instructions are not the same thing as enforcement. Telling an agent not to edit a protected file is weaker than placing a deterministic gate in front of the write operation. Hooks give teams a place to attach existing policy, security and quality checks without trying to teach every rule to the model. ## Token budgets finally put a ceiling on autonomous loops Managed agents can be given a `max_total_tokens` budget covering input, output and thinking tokens. When a run reaches that limit, execution returns incomplete while preserving its environment and context so the developer can inspect or continue it with a fresh budget. Google notes that the limit is best-effort and actual usage can slightly exceed it between budget checks. This is a small control with a large practical effect. Autonomous agents can take unexpected paths, repeat failed approaches or spend far more than a short interactive prompt. A hard budget makes cost part of the runtime contract rather than something discovered after the invoice arrives. ## Scheduled triggers move the product toward persistent workers Google has also added scheduled triggers that bind an agent, prompt, environment and cron schedule together. Runs reuse the same sandbox, so files can persist between executions. That makes Managed Agents relevant to recurring jobs such as dependency audits, periodic research, repository maintenance or report generation. Builders should still think carefully about idempotency, stale state and what happens when one scheduled run overlaps another, but the platform is clearly moving beyond request-response agents. ## Gemini 3.7 Flash is now the default — pin the model when stability matters Google's July 28 announcement made Gemini 3.6 Flash the default model for the `antigravity-preview-05-2026` runtime. The current Antigravity Agent documentation, last updated August 13, now says the same agent defaults to `gemini-3.7-flash` when `agent_config.model` is omitted. Gemini 3.6 Flash, Gemini 3.5 Flash and Gemini 3.5 Flash-Lite remain selectable. For builders, the important point is not just that the newer model is stronger. It is that the underlying default can change while the managed-agent identifier stays the same. Teams that need reproducible behavior should explicitly set the model and treat a model upgrade like any other production dependency change: evaluate it on real traces, then roll it out deliberately. ## The default switch does not currently change the published token rate Antigravity Agent pricing is pay-as-you-go based on the underlying Gemini model tokens and the tools the agent uses. Google's current Gemini API pricing lists the same promotional standard rate for Gemini 3.7 Flash and Gemini 3.6 Flash through December 31, 2026: $0.75 per million input tokens and $3.75 per million output tokens, with published rates of $1.50 input and $7.50 output starting January 1, 2027. The equal headline rate means the default move to 3.7 Flash does not by itself increase today's per-token price. Total task cost can still change if the newer model uses a different number of reasoning tokens, tool calls or retries, so production comparisons should measure end-to-end job cost rather than assuming equal token rates mean equal bills. ## Key details - The Antigravity managed-agent runtime now defaults to Gemini 3.7 Flash when `agent_config.model` is omitted; the current documentation was updated August 13, 2026. - Gemini 3.6 Flash remains available as an explicit model choice alongside Gemini 3.5 Flash and Gemini 3.5 Flash-Lite. - Gemini 3.7 Flash and 3.6 Flash currently share the same promotional standard API rate through December 31, 2026: $0.75/1M input tokens and $3.75/1M output tokens, rising to $1.50/$7.50 on January 1, 2027. - Environment hooks can run before or after tool calls and can deny actions or run custom validation. - A `max_total_tokens` control can pause an autonomous run when it reaches a defined token budget while preserving state; Google says enforcement is best-effort and can slightly overshoot between checks. - Scheduled triggers can run recurring agent jobs and reuse the same sandbox between executions. - Managed Agents remain in preview and are available to both free-tier and paid-tier Gemini API projects. - The Environments API lets developers inspect and clean up remote sandbox sessions. ## Builder takeaways - Pin `agent_config.model` for production workloads where reproducibility matters. The default behind `antigravity-preview-05-2026` has already moved from Gemini 3.6 Flash to 3.7 Flash without a new agent identifier. - Re-run your own agent traces before adopting a new default model. Compare task success, retries, tool-call count, wall-clock latency, token use and total job cost rather than assuming a newer default is automatically better for your workflow. - Use hooks for rules that must actually be enforced. Keep security gates, linting and protected-file checks outside the model's discretion. - Set explicit token budgets on long-running jobs before treating them as unattended workers, and leave margin for the documented best-effort budget enforcement. - Treat persistent sandboxes as stateful infrastructure: design for stale files, retries, overlapping runs and cleanup. - Model costs beyond December 31, 2026 as well as today's promotional rate; the published standard Gemini 3.7 Flash and 3.6 Flash rates double on January 1, 2027. - Start scheduled agents on bounded, reversible jobs before giving them write access to production systems. ## What to watch - Whether Google moves Managed Agents out of preview and publishes stronger production guarantees. - Whether the implicit Antigravity default changes again without a new agent identifier, and how much notice Google gives builders before future default-model transitions. - How pricing and total task cost behave after the Gemini 3.7/3.6 promotional rates expire on December 31, 2026. - Whether hooks expand into richer policy and approval primitives. - How developers handle overlapping scheduled runs and long-lived sandbox state in practice. - Whether the runtime remains tightly coupled to Gemini or becomes more flexible around external models and agent harnesses. ## Uncertainties - Google's launch material and current documentation demonstrate the controls, but there is still not enough public production data to judge reliability across large numbers of long-running scheduled agents. - Persistent environments reduce setup work but also create operational questions around stale state, secrets and cleanup that each application still has to design for. - Equal per-token pricing does not imply equal end-to-end job cost: Gemini 3.7 Flash may use a different mix of reasoning tokens, tool calls and retries than 3.6 Flash on a given workload. - The Antigravity Agent and Interactions API remain in preview, and Google explicitly notes that features and schemas may change. ## Timeline - **2026-05-19 — Google launches Managed Agents:** The Gemini API gains a managed Antigravity agent running in an isolated Linux environment. - **2026-07-07 — Background work and remote MCP arrive:** Google adds background execution, remote MCP support, custom functions and credential refresh. - **2026-07-28 — Budgets, hooks and schedules make the runtime more operational:** Managed Agents switch to Gemini 3.6 Flash by default and gain environment hooks, token budgets, scheduled triggers and free-tier access. - **2026-08-13 — Gemini 3.7 Flash becomes the managed-agent default:** Google's updated Antigravity Agent documentation says `antigravity-preview-05-2026` now defaults to Gemini 3.7 Flash when no model is explicitly selected; 3.6 Flash remains supported. - **2026-12-31 — Promotional Flash pricing ends:** Google's published promotional standard rate for Gemini 3.7 Flash and 3.6 Flash runs through this date. - **2027-01-01 — Published standard Flash pricing doubles:** Google lists $1.50/1M input tokens and $7.50/1M output tokens for Gemini 3.7 Flash and 3.6 Flash starting on this date. ## Sources - [Gemini API Managed Agents: 3.6 Flash, hooks, and more](https://blog.google/innovation-and-ai/technology/developers-tools/expanding-managed-agents-gemini-api-3-6-flash-hooks/) — Google · Primary announcement · 2026-07-28T00:00:00+12:00. Original announcement for the 3.6 Flash default, environment hooks, budgets, scheduled triggers, free-tier access and Environments API. - [Antigravity Agent](https://ai.google.dev/gemini-api/docs/antigravity-agent) — Google AI for Developers · Official documentation · 2026-08-13T00:00:00+12:00. Current managed-agent documentation. It states that `antigravity-preview-05-2026` is powered by and defaults to Gemini 3.7 Flash, lists supported model choices, preview status, budget semantics, triggers and availability. - [Gemini Developer API pricing](https://ai.google.dev/gemini-api/docs/pricing) — Google AI for Developers · Official pricing documentation. Current token pricing for Gemini 3.7 Flash and 3.6 Flash, including promotional rates through December 31, 2026 and published rates from January 1, 2027. - [Introducing Gemini 3.7 Flash](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-3-7-flash/) — Google · Primary model announcement · 2026-08-13T00:00:00+12:00. Primary launch announcement for Gemini 3.7 Flash and its positioning for coding and agent workflows. - [Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-6-flash-3-5-flash-lite-3-5-flash-cyber/) — Google · Primary model announcement · 2026-07-21T00:00:00+12:00. 3.6 Flash positioning, efficiency claims, availability and pricing context. - [Introducing Managed Agents in the Gemini API](https://blog.google/innovation-and-ai/technology/developers-tools/managed-agents-gemini-api/) — Google · Primary technical announcement · 2026-05-19T00:00:00+12:00. Original managed-agent architecture and isolated Linux environment.