# Confidence Comes from Experience: Experiential Confidence Estimation from Reasoning to Agents

arXiv: https://arxiv.org/abs/2609.17708 · HTML: https://mlsanity.com/paper/2609.17708/

## Key facts

- No citing papers in the map yet (data through 7 Oct 2026).
- Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning.
- Builds on 21 of its 74 references that are in the map.
- First posted on arXiv 15 Sep 2026.

## Placed in

- LLM reasoning / [Test-time compute and efficient thinking](https://mlsanity.com/lr/test-time-compute-and-efficient-thinking/)
- LLM agents and tool use / [General LLM agent frameworks and benchmarks](https://mlsanity.com/agents/general-llm-agent-frameworks-and-benchmarks/)

## Builds on (most cited first, 21 of 21)

- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [Language Models (Mostly) Know What They Know](https://mlsanity.com/paper/2207.05221/) (2022)
- [Teaching Models to Express Their Uncertainty in Words](https://mlsanity.com/paper/2205.14334/) (2022)
- [CodeBLEU: a Method for Automatic Evaluation of Code Synthesis](https://mlsanity.com/paper/2009.10297/) (2020)
- [ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory](https://mlsanity.com/paper/2509.25140/) (2025)
- [Conformal Prediction with Large Language Models for Multi-Choice Question Answering](https://mlsanity.com/paper/2305.18404/) (2023)
- [R-Tuning: Instructing Large Language Models to Say `I Don't Know'](https://mlsanity.com/paper/2311.09677/) (2023)
- [SaySelf: Teaching LLMs to Express Confidence with Self-Reflective Rationales](https://mlsanity.com/paper/2405.20974/) (2024)
- [Gemini Embedding: Generalizable Embeddings from Gemini](https://mlsanity.com/paper/2503.07891/) (2025)
- [Can Large Language Models Faithfully Express Their Intrinsic Uncertainty in Words?](https://mlsanity.com/paper/2405.16908/) (2024)
- [Linguistic Calibration of Long-Form Generations](https://mlsanity.com/paper/2404.00474/) (2024)
- [How do LLMs Compute Verbal Confidence](https://mlsanity.com/paper/2603.17839/) (2026)
- [Agentic Uncertainty Quantification](https://mlsanity.com/paper/2601.15703/) (2026)
- [Beyond the Final Layer: Intermediate Representations for Better Multilingual Calibration in Large Language Models](https://mlsanity.com/paper/2510.03136/) (2025)
- [Causal Evidence that Language Models use Confidence to Drive Behavior](https://mlsanity.com/paper/2603.22161/) (2026)
- [Reported Confidence in LLMs Tracks Commitment More Than Correctness](https://mlsanity.com/paper/2606.29490/) (2026)
- [Generalized Correctness Models: Learning Calibrated and Model-Agnostic Correctness Predictors from Historical Patterns](https://mlsanity.com/paper/2509.24988/) (2025)
- [Code Is More Than Text: Uncertainty Estimation for Code Generation](https://mlsanity.com/paper/2606.09577/) (2026)
- [How LLMs Detect and Correct Their Own Errors: The Role of Internal Confidence Signals](https://mlsanity.com/paper/2604.22271/) (2026)
- [Critic Experience Bank: Self-Evolving Step-Level Confidence Estimation for LLM Agents](https://mlsanity.com/paper/2607.12397/) (2026)
- [The Computational Basis of Confidence in Large Language Models](https://mlsanity.com/paper/2607.12447/) (2026)

## Abstract (arXiv)

Reliable confidence estimation is increasingly central to the trustworthy deployment of language models: a calibrated estimate of the probability that an output is correct decides what to ship, what to escalate, and what to retry. Existing confidence estimators, however, share one design premise: they only read the current inference process, either by introspecting on it, scoring its token probabilities, or resampling it. We argue that the current inference is not a sufficient basis for confidence. We propose XConf (eXperiential Confidence): estimating confidence together with the model's accumulated experience. The experience is stored as a record of the model's own graded past episodes, each holding the task, the model's reflection, its stated confidence, the outcome, and a lesson written once the grade arrived. Given a new task, XConf's Recall stage retrieves past episodes on similar tasks met with a similar stated confidence, and reads off their historical success rate; its Reflect stage shows the model this record, has it name its recurring failure mode, and restate a confidence now informed by its own track records. Our estimator is format-general, requiring no logit access or weight updates, and costs only one answer generation. Across nine benchmarks spanning reasoning, coding, multimodal QA, and interactive agents, and four models from three families, XConf beats or matches ten-sample self-consistency in discrimination (AUROC) on 23 of 24 comparisons, with much lower calibration error (ECE), at a tenth of the generation cost. Used for selective prediction, abstaining on the 10% least-confident episodes raises the delivered success rate by up to 8.7 points on agent tasks. We therefore see experiential confidence estimation as a new paradigm for future general-purpose confidence estimation.