Both trace agents, run experiments, score production traces and route traces to human reviewers, and both now belong to infrastructure companies. Galileo, now Splunk Agent Observability, has the broader run-time checks, applied to tool inputs and outputs from central rules; Weave has the stronger prompt tooling and a free self-serve start, next to W&B model training.[3][6]
The main difference between Galileo and W&B Weave is how they act during a run: Galileo's Agent Control applies central rules to model and tool inputs and outputs without code changes, while Weave's guardrails are scorers you call from Python code. Both are now sold inside larger platforms: Cisco completed its acquisition of Galileo in May 2026 and has sold it as Splunk Agent Observability since 7 August 2026, and CoreWeave, which bought W&B in May 2025, has sold Weave inside CoreWeave Forge since September 2026. Weave has a free tier; Splunk lists none for this product.
| Decision factor | Galileo | W&B Weave |
|---|---|---|
| Ownership | Cisco (completed May 2026); sold as Splunk Agent Observability since 7 August 2026 | CoreWeave since May 2025; sold inside CoreWeave Forge since 30 September 2026 |
| Runtime guardrails | Yes: Agent Control checks model and tool inputs and outputs during a run, no code changes | Partial: scorers called from Python code can block or modify inputs and outputs |
| Evaluators | LLM judges, Luna-2 small models, Python code metrics (not in the Splunk-hosted edition) | LLM judges, predefined scorers, Python and TypeScript function scorers |
| Prompt tooling | Prompts and playground; version control not documented | Versioned prompts and a playground |
| Self-hosting | On-premises, including air-gapped (commercial) | Kubernetes with a Weave licence and a W&B Platform install |
| Free tier | None listed for Agent Observability | Forge Free: 1 GB of Weave ingestion a month |
| Paid pricing | From $100 a month per 1.2 million spans a year, billed annually | Forge Pro: from $60 a month, 1.5 GB of Weave ingestion |
| Capability | Galileo | W&B Weave | Prefactor |
|---|---|---|---|
| Observe | |||
| LLM call tracing | ✓ | ✓ | ✓ |
| Multi-step agent tracing (tool calls, nested spans) | ✓ | ✓ | ✓ |
| Cost tracking | ✓ | ✓ | ✓ |
| OpenTelemetry support | ✓ | ✓ | ✓ |
| Evaluate | |||
| Offline evals (datasets, experiments) | ✓ | ✓ | ✓ |
| Online evals on production traffic | ✓ | ✓ | ✓ |
| LLM-as-a-judge scoring | ✓ | ✓ | ✓ |
| Code-based evaluators | ✓ | ✓ | ✓ |
| Human review and annotation | ✓ | ✓ | ✓ |
| Prompt versioning and playground | Partial | ✓ | — |
| Act on findings | |||
| Runtime controls on live agent actions | ✓ | Partial | ✓ |
| Gateway or proxy in the request path | — | Not stated | — |
| Licence and deployment | |||
| Open source | Partial | Partial | Partial |
| Self-hosting | ✓ | ✓ | Partial |
Cisco completed its acquisition of Galileo in May 2026, and since 7 August 2026 the product has been sold as Splunk Agent Observability inside Splunk Observability Cloud. Customers who joined before that date still use Galileo's documentation; later customers use Splunk's. The old pricing page now redirects to Splunk.
CoreWeave completed its acquisition of Weights & Biases in May 2025, and on 30 September 2026 launched CoreWeave Forge, which bundles W&B Models, Weave and new CoreWeave tools. W&B hosted accounts now sign in through Forge, and Weave's pricing, docs and login have moved to CoreWeave domains. CoreWeave also recommends Agent Lens, a separate agent product in public preview, for agent tracing.
The parent platforms point at different buyers. Splunk is an observability and operations platform, so Galileo suits teams that already send infrastructure telemetry there. Forge is a model training and inference platform, so Weave suits teams that train or fine-tune models with W&B. Both products are still settling into their new homes, so expect packaging changes over the next few quarters.
This is the clearest functional difference. Galileo's Agent Control evaluates model and tool inputs and outputs while an agent runs, and can block prompt injection, personal data leaks and harmful content using rules set in one place, without changes to agent code. It replaced the older Protect product in June 2026, and its core was released as open source under Apache-2.0 in March 2026, so it can be self-hosted from its repository.
Weave's guardrails are scorers you call from Python application code. They can block or modify inputs and outputs before they reach users, but each check is wired into your code, the TypeScript SDK does not support them, and Weave documents no checks on tool actions. Neither product documents approval workflows for agent actions.
Galileo offers preset and custom LLM judges, Luna-2 small models as lower-cost judges and Python code metrics. One limit to check: the Splunk-hosted edition supports five preset Luna evaluators and no custom code evaluators; those are available on-premises only. Galileo's prompts can be stored, templated and run in a playground, but its prompt documentation does not describe version control.
Weave scorers can be Python or TypeScript functions, or classes in Python, and it adds local small-model scorers for toxicity and context relevance. Prompts are versioned objects, and the playground compares models side by side and tests prompts against stored traces. Both products have annotation queues for human review. If prompt iteration is a daily job, Weave covers more of it.
Both have Python and TypeScript SDKs. Galileo added multi-turn experiments in June 2026 and offers Autotune, which adjusts judge metrics from reviewer feedback. Weave's Evaluation API compares runs to catch regressions and shares results as leaderboards, and its EvaluationLogger lets you log evaluations from existing code without restructuring it.
As of 20 October 2026, Splunk lists Agent Observability from $100 a month per 1.2 million spans a year, billed annually, including 500 million Luna tokens and five Luna fine-tuning runs a year; no free tier is listed for this product. CoreWeave prices Weave through Forge: Forge Free includes 1 GB of Weave ingestion a month and up to 5 model seats, and Forge Pro starts at $60 a month, billed monthly, with 1.5 GB for teams under 50 employees and extra ingestion listed at $0.10 per MB.
Weave can be tried today without a sales call; Galileo is an annual purchase through Splunk. For deployment, Splunk offers Agent Observability on-premises, including air-gapped, while self-hosted Weave needs a Weave-enabled licence and an existing W&B Platform install. Check Splunk's pricing and CoreWeave Forge pricing before committing.
We sell the layer this section describes. Read it with that in mind.
Stated plainly, since Prefactor publishes this page: Prefactor is an agent evaluation platform that scores each step of a run and acts on what it finds through runtime policies on tool calls and data access, approval holds for risky actions and a kill switch for a run, an agent or a team. Of these two, Galileo's Agent Control is the closer match on blocking; Weave's guardrails check content from Python code. Neither documents holding an action for a person to approve. Prefactor is self-serve and not tied to Splunk or CoreWeave, but it has no prompt playground or prompt registry and its platform is hosted. See the full comparison of 15 platforms.
Reviewed against public product and documentation pages on October 20, 2026. If a vendor has changed a feature, product name, or positioning since then, send a correction and we will update it. Numbered source links in the page body point to the ordered sources below.
Prefactor scores each step of every agent run and can hold or stop an action when a score says it should not happen. Start on the free tier: 5,000 spans a month, no credit card required.
Prefactor helps teams observe, evaluate, and improve their AI agents in production, across every framework and provider.