v0.11.0

What's Changed

• Bump to v0.10.1 by @rhatdan in #1667
• Adds the ability to include vision based context to chat via --rag by @ieaves in #1661
• Update registry.access.redhat.com/ubi9/ubi Docker tag to v9.6-1751897624 by @red-hat-konflux-kflux-prd-rh03[bot] in #1670
• chore(deps): update konflux references by @red-hat-konflux-kflux-prd-rh03[bot] in #1664
• feat: allow for dynamic version installing of ramalama-stack by @nathan-weinberg in #1671
• Inspect add safetensor support by @engelmi in #1666
• Revert "feat: allow for dynamic version installing of ramalama-stack" by @ericcurtin in #1672
• move --image & --keep-groups to run, serve, perplexity, bench commands by @rhatdan in #1669
• mlx fixes by @ericcurtin in #1673
• Enhance ref file and mount all snapshot files to container by @engelmi in #1643
• Hide --container option, having --container/--nocontainer is confusing by @rhatdan in #1675
• Enable SELinux separation by @rhatdan in #1676
• chore: bump ramalama-stack to 0.2.5 by @nathan-weinberg in #1680
• Bugfix for chat by @ericcurtin in #1679
• Update registry.access.redhat.com/ubi9/ubi Docker tag to v9.6-1752069608 by @renovate[bot] in #1668
• chore(deps): update registry.access.redhat.com/ubi9/ubi docker tag to v9.6-1752069608 by @red-hat-konflux-kflux-prd-rh03[bot] in #1684
• konflux: add integration tests that run in multi-arch VMs by @mikebonnet in #1683
• Allow ramalama rag to output different formats by @rhatdan in #1685
• Bug/chat fix by @ieaves in #1681
• chore(deps): update konflux references by @red-hat-konflux-kflux-prd-rh03[bot] in #1688
• Only install if pyproject.toml exists by @ericcurtin in #1689
• Readme improvements: Update model's name and improve CUDA_VISIBLE_DEVICES section by @mbortoli in #1691
• Move rpms by @smooge in #1693

New Contributors

• @mbortoli made their first contribution in #1691

Full Changelog: v0.10.1...v0.11.0

v0.10.1

What's Changed

• Bump to v0.10.0 by @rhatdan in #1629
• Fix handling of --host option when running in a container by @rhatdan in #1628
• Start process of moving python-ramalama to ramalama by @smooge in #1498
• Fix modelstore deleting logic when multiple reference refer to the same blob/snapshot by @olliewalsh in #1620
• chore(deps): update registry.access.redhat.com/ubi9/ubi docker tag to v9.6-1751287003 by @red-hat-konflux-kflux-prd-rh03 in #1633
• run tests during build pipelines by @mikebonnet in #1614
• Split the model store into multiple files by @engelmi in #1640
• chore: bump ramalama-stack to 0.2.4 by @nathan-weinberg in #1639
• Use config instance for defining pull behavior in accel_image by @engelmi in #1638
• quadlet: add missing privileged options by @jbtrystram in #1631
• build layered images from Containerfiles by @mikebonnet in #1641
• Add command to list available models by @ericcurtin in #1635
• Adds a user configuration setting to disable gpu prompting by @ieaves in #1632
• chore(deps): update registry.access.redhat.com/ubi9/ubi docker tag to v9.6-1751445649 by @red-hat-konflux-kflux-prd-rh03 in #1658
• Update lint and format tools configuration by @telemaco in #1659
• konflux: add pipelines for the layered images of ramalama, cuda, rocm, and rocm-ubi by @mikebonnet in #1657
• Always use absolute path for --store option by @rhatdan in #1637
• Add .pre-commit-config.yaml by @telemaco in #1660
• MLX runtime support by @kush-gupt in #1642
• Make sure errors and progress messages go to STDERR by @rhatdan in #1665

New Contributors

• @jbtrystram made their first contribution in #1631
• @telemaco made their first contribution in #1659

Full Changelog: v0.10.0...v0.10.1

v0.10.0

What's Changed

• Bump to v0.9.3 by @rhatdan in #1586
• Remove last libexec program by @rhatdan in #1576
• Don't pull image when doing ramalama --help call by @rhatdan in #1589
• API key support by @ericcurtin in #1578
• Move RamaLama container image to default to fedora:42 by @rhatdan in #1595
• Missing options of api_key and pid2kill are causing crashes by @rhatdan in #1601
• Some of our tests are running for hours, need to be timed out by @rhatdan in #1602
• chore(deps): update registry.access.redhat.com/ubi9/ubi docker tag to v9.6-1750786174 by @red-hat-konflux-kflux-prd-rh03 in #1600
• konflux: centralize pipeline definitions by @mikebonnet in #1599
• Allow std input by @ericcurtin in #1606
• konflux: use shared pipelines for rocm, rocm-ubi, and cuda by @mikebonnet in #1608
• Prune model store code by @engelmi in #1607
• Switchout hasattr for getattr wherever possible by @rhatdan in #1605
• add support for running bats in a container by @mikebonnet in #1598
• Separate build image into its own VM by @rhatdan in #1609
• container-images: pin mesa version to COPR by @slp in #1603
• konflux: build bats image by @red-hat-konflux-kflux-prd-rh03 in #1612
• rename "nopull" boolean to "pull" by @ktdreyer in #1611
• Use standard zsh completion directory by @carlwgeorge in #1619
• Free up disk space for building all images by @rhatdan in #1615
• Fix removing of file based URL models by @rhatdan in #1610
• chore: bump ramalama-stack to 0.2.3 by @nathan-weinberg in #1616
• Fixup to work with llama-stack by @rhatdan in #1588
• Fix unit tests for machines with GPUs by @sarroutbi in #1621
• Want to pick up support for gemma3n by @ericcurtin in #1623
• Add gemma aliases by @ericcurtin in #1624
• Adds the ability to pass files to ramalama run by @ieaves in #1570
• chore(deps): update konflux references by @red-hat-konflux-kflux-prd-rh03 in #1627

New Contributors

• @ktdreyer made their first contribution in #1611
• @carlwgeorge made their first contribution in #1619

Full Changelog: v0.9.3...v0.10.0

v0.9.3

What's Changed

• Convert tabs to spaces by @ericcurtin in #1538
• Make minimum version of Python consistent by @Hasnep in #1512
• Upgrade podman by @ericcurtin in #1540
• Bump to v0.9.2 by @rhatdan in #1537
• Downgrade whisper by @ericcurtin in #1543
• Deduplicate code by @ericcurtin in #1539
• Add dnf update -y to Fedora ROCm build by @ericcurtin in #1544
• model: always pass in GPU offloading parameters by @alaviss in #1502
• Run bats test with TMPDIR pointing at /mnt/tmp by @rhatdan in #1548
• Tabs to spaces by @ericcurtin in #1549
• Add GGML_VK_VISIBLE_DEVICES env var by @ericcurtin in #1547
• Create tempdir when run as non-root user by @rhatdan in #1551
• Red Hat Konflux kflux-prd-rh03 update ramalama by @red-hat-konflux-kflux-prd-rh03 in #1542
• chore(deps): update registry.access.redhat.com/ubi9/ubi docker tag to v9.6-1749542372 by @red-hat-konflux-kflux-prd-rh03 in #1555
• Fix default prefix for systems with no engines by @rhatdan in #1556
• Add install command via homebrew by @scraly in #1558
• Remove Model flag for safetensor files for now by @engelmi in #1559
• Add verbose rule for complete output on unit tests by @sarroutbi in #1562
• Reuse code for unit test execution rules by @sarroutbi in #1564
• :latest tag should not be assumed for non-OCI artefacts by @ericcurtin in #1534
• Replace ramalama-client-code with ramalama chat by @rhatdan in #1550
• Document the image format created/consumed by the oci:// transport by @mtrmac in #1569
• Trying to save space by @ericcurtin in #1541
• Fix test_accel unit test to fallback to latest by @sarroutbi in #1567
• install ramalama into containers from the current checkout by @mikebonnet in #1566
• TMT: run tests with GPUs by @lsm5 in #1101
• fix: vLLM serving and model mounting by @kush-gupt in #1571
• Make model argument mandatory by @ericcurtin in #1574
• fix: broken link in CI dashboard by @nathan-weinberg in #1580
• chore: bump ramalama-stack to 0.2.2 by @nathan-weinberg in #1579

New Contributors

• @red-hat-konflux-kflux-prd-rh03 made their first contribution in #1542
• @scraly made their first contribution in #1558
• @mtrmac made their first contribution in #1569

Full Changelog: v0.9.2...v0.9.3

v0.9.2

What's Changed

• Only print this in the llama-stack case by @ericcurtin in #1486
• Throw exception when using OCI without engine by @rhatdan in #1471
• Make sure llama-stack URL is shown to user by @rhatdan in #1490
• Fix #1489 by @yeahdongcn in #1491
• There's a change that we want that avoids using software rasterizers by @ericcurtin in #1495
• Install uv to fix build issue by @ericcurtin in #1496
• chore(deps): update registry.access.redhat.com/ubi9/ubi docker tag to v9.6-1749542372 by @renovate in #1492
• Only enumerate ROCm-capable AMD GPUs by @alaviss in #1500
• amdkfd: add constants for heap types by @alaviss in #1501
• This is not a multi-model model by @ericcurtin in #1499
• fix: remove unneeded dependency from Llama Stack container by @nathan-weinberg in #1503
• Increase retry attempts to attempt to connect to server by @ericcurtin in #1507
• Ignore errors when removing snapshot directory by @engelmi in #1511
• Add Python shebang files to linting by @Hasnep in #1514
• For ramalama ls shorten huggingface lines by @ericcurtin in #1516
• Update black target version by @Hasnep in #1513
• Wait for upto 16 seconds for model to load by @ericcurtin in #1510
• This installs ramalama via uv if python3 version is too old by @ericcurtin in #1497
• fix(deps): update dependency huggingface-hub to ~=0.33.0 by @renovate in #1505
• chore(common/intel_gpus): detect arc a770, a750 by @kwaa in #1517
• Do not run with --tty when not in interactive mode by @rhatdan in #1506
• Update to add multi-modal by @rhatdan in #1522
• Add --all option to ramalama ls by @engelmi in #1528
• Add colors to "ramalama serve" if we can by @ericcurtin in #1529
• Change the FROM for asahi container image by @ericcurtin in #1523
• Refactor config and arg typing by @ieaves in #1488
• Add ramalama chat command by @rhatdan in #1531
• Suggest using uv pip install to get missing module by @rhatdan in #1532
• Not sure this is supposed to be here by @ericcurtin in #1535
• chore: bump ramalama-stack to 0.2.1 by @nathan-weinberg in #1536
• honor the user specifying the image by @rhatdan in #1527

New Contributors

• @kwaa made their first contribution in #1517

Full Changelog: v0.9.1...v0.9.2

v0.9.1

What's Changed

• feat: s390x build commands by @taronaeo in #1459
• docs: update container_build.sh help information by @taronaeo in #1461
• chore: remove unclear else from llama and whisper build by @taronaeo in #1464
• fix: lock down ramalama-stack version in llama-stack Containerfile by @nathan-weinberg in #1465
• Rename: RepoFile=>HFStyleRepoFile, BaseRepository=>HFStyleRepository, BaseRepoModel=>HFStyleRepoModel by @yeahdongcn in #1466
• Documentation improvements by @waltdisgrace in #1468
• Change timeouts by @ericcurtin in #1469
• llama-stack container build fails with == 1.5.0 by @rhatdan in #1467
• Do not override a small subset of env vars by @ericcurtin in #1475
• Call set_gpu_type_env_vars rather than set_accel_env_vars by @ericcurtin in #1476
• Don't warmup by default by @ericcurtin in #1477
• chore: bump 'ramalama-stack' version to 0.2.0 by @nathan-weinberg in #1478
• Adds dev dependency groups by @ieaves in #1481
• fix(deps): update dependency huggingface-hub to ~=0.32.4 by @renovate in #1483
• Fix handling of generate with llama-stack by @rhatdan in #1472
• Update demos to show serving models. by @rhatdan in #1474
• Bump to v0.9.1 by @rhatdan in #1484

New Contributors

• @waltdisgrace made their first contribution in #1468

Full Changelog: v0.9.0...v0.9.1

v0.9.0

What's Changed

• chore: bump llama.cpp to support tool streaming by @p5 in #1438
• Bump to v0.8.5 by @rhatdan in #1439
• fix: update references to Python 3.8 to Python 3.11 by @nathan-weinberg in #1441
• Fix quadlet handling of duplicate options by @olliewalsh in #1442
• fix(gguf_parser): fix big endian model parsing by @taronaeo in #1444
• Choice could be not set and should not be used by @rhatdan in #1447
• fix(run): Ensure 'run' subcommand works with host proxy settings. by @melodyliu1986 in #1430
• Switch default ramalama image build to use VULKAN by @rhatdan in #1449
• make ramalama-client-core send default model to server by @rhatdan in #1450
• fix(gguf_parser): fix memoryerror exception when loading non-native models by @taronaeo in #1452
• Small logging improvements by @almusil in #1455
• feat(model_store): prevent model endianness mismatch on download by @taronaeo in #1454
• Add support for llama-stack by @rhatdan in #1413
• Refactoring huggingface.py and modelscope.py and extract repo_model_base.py by @yeahdongcn in #1456
• Eliminate selinux-policy packages from containers by @rhatdan in #1451
• Snapshot verification by @engelmi in #1458
• Add support for generating kube.yaml and quadlet/kube files for llama… by @rhatdan in #1457
• Bump to v0.9.0 by @rhatdan in #1462

New Contributors

• @taronaeo made their first contribution in #1444
• @almusil made their first contribution in #1455

Full Changelog: v0.8.5...v0.9.0

v0.8.5

What's Changed

• Shorten url in README.md by @ericcurtin in #1392
• Bump to 0.8.3 by @rhatdan in #1391
• This script is not macOS only by @ericcurtin in #1393
• Using perror in cli.main by @ieaves in #1395
• Fix builds by @ericcurtin in #1396
• Remove all path additions to this file by @ericcurtin in #1398
• Fix issues reported by pylint for cli.py by @sarroutbi in #1402
• Increase cli.py coverage by @sarroutbi in #1403
• Add minor CONTRIBUTING.md enhancements by @sarroutbi in #1404
• Include additional information in CONTRIBUTING.md by @sarroutbi in #1406
• Fix cuda builds installation of python3.11 by @rhatdan in #1399
• added a docling ocr flag ( text image recognition) flag to address RAM issue by @bmahabirbu in #1400
• fix: removed ocr print statement and updated ocr description by @bmahabirbu in #1408
• Support Moore Threads GPU #2 by @yeahdongcn in #1410
• Add more debug for non starting servers with "ramalama run" by @ericcurtin in #1415
• Small typo by @ericcurtin in #1418
• Multimodal/vision support by @olliewalsh in #1416
• Added host:container port mapping to quadlet generation by @engelmi in #1409
• Don't throw Exceptions, be more specific by @rhatdan in #1420
• Normalize hf repo quant/tag by @olliewalsh in #1422
• Support Moore Threads GPU #1 by @yeahdongcn in #1407
• Add smolvlm vision models by @ericcurtin in #1424
• Update registry.access.redhat.com/ubi9/ubi Docker tag to v9.6-1747219013 by @renovate in #1423
• Bump llama.cpp to fix rocm bug by @afazekas in #1427
• Remove unused parameters from ollama_repo_utils.py by @sarroutbi in #1428
• Add support for Hugging Face token authentication by @olliewalsh in #1425
• split/big model support for llama.cpp by @afazekas in #1426
• Don't use jinja in the multimodal case by @ericcurtin in #1435
• Support Moore Threads GPU #3 by @yeahdongcn in #1436

New Contributors

• @olliewalsh made their first contribution in #1416

Full Changelog: v0.8.3...v0.8.5

v0.8.3

What's Changed

• Bump to v0.8.2 by @rhatdan in #1346
• Fix typo (RAMALAMA_TRANSPORTS->RAMALAMA_TRANSPORT) by @sarroutbi in #1352
• Fix partial model listing by @engelmi in #1350
• Consolidate and alphabetize runtime options by @rhatdan in #1349
• Use CLI ramalama serve here by @ericcurtin in #1345
• Use model type instead of class name by @engelmi in #1353
• Add coverage tools, run them via specific rules by @sarroutbi in #1358
• Add TOMLParser unit tests by @sarroutbi in #1356
• chore: update curl commands in llama-stack Containerfile by @nathan-weinberg in #1360
• Allow user-defined CUDA_VISIBLE_DEVICES environment variable by @mcornea in #1355
• Groom coverage rules, genreate xml/lcov reports by @sarroutbi in #1365
• Extend TOMLParser coverage to 100% by @sarroutbi in #1364
• small improvements to the build of the ramalama-rag image by @mikebonnet in #1361
• Allow docling to handle URLs rather then handling locally by @rhatdan in #1359
• fix "ramalama client" by @mikebonnet in #1366
• Use all GPUs in CUDA_VISIBLE_DEVICES as default by @mcornea in #1369
• Use python3.11 on systems with older python by @rhatdan in #1367
• Update not found urls by @jelly in #1370
• Various CI fixes by @nathan-weinberg in #1372
• Make version optional in build by @rhatdan in #1373
• Remove hardcoded /usr/local site-packages injection to fix sys.path pollution by @ieaves in #1340
• Update llama_cpp_sha to the latest version by @TristanCacqueray in #1378
• ci: modify 'latest' job to only run on release by @nathan-weinberg in #1375
• ci: additional fixes and cleanup for image build jobs by @nathan-weinberg in #1384
• Check nvidia-container-runtime executable also in engine.py by @antbbn in #1380
• Support older version of Docker by @yeahdongcn in #1383
• Punctuation consistency when pulling models by @ericcurtin in #1386
• ci(fix): macOS runner didn't have bats by @nathan-weinberg in #1388
• Fix #1382 by @yeahdongcn in #1387
• update the shortnames path according to the shortnames.py by @melodyliu1986 in #1363
• Add support for modelscope and update doc by @yeahdongcn in #1381
• Add output path to generate quadlet/kube by @engelmi in #1371
• More de-duplication and consistency by @ericcurtin in #1389

New Contributors

• @mcornea made their first contribution in #1355
• @mikebonnet made their first contribution in #1361
• @jelly made their first contribution in #1370
• @ieaves made their first contribution in #1340
• @antbbn made their first contribution in #1380
• @yeahdongcn made their first contribution in #1383

Full Changelog: v0.8.2...v0.8.3

v0.8.2

What's Changed

• Avoid unused parameter by @sarroutbi in #1309
• Pass args to ramalama run core by @ericcurtin in #1292
• common: adjust nvidia-smi for check cuda version by @dougsland in #1311
• Simplify installer by @ericcurtin in #1312
• feat: update llama-stack Containerfile to use ramalama-stack by @nathan-weinberg in #1314
• Fix up several issue in llama-stack Containerfile by @rhatdan in #1317
• Updates to ReadMe doc by @arburka in #1319
• Update Docs page by @arburka in #1320
• Switch all Ramalama to RamaLama by @rhatdan in #1323
• Turn on client/server implementation of run by @ericcurtin in #1158
• Expose http line in man pages by @rhatdan in #1327
• fix: additional fixes for llama-stack Containerfile by @nathan-weinberg in #1330
• Fixup use of /.venv by @rhatdan in #1331
• Add shortnames for mistral-small3.1 model by @ericcurtin in #1329
• Fixed mac gpu not being enabled from stale global var check by @bmahabirbu in #1333
• Make installer more resilliant by @ericcurtin in #1332
• RamaLamaShell fixes by @ericcurtin in #1334
• Describe what this test does by @ericcurtin in #1328
• llama stack run should be the CMD not run during build by @rhatdan in #1335
• INFERENCE_MODEL should be set by the container engine by @rhatdan in #1336
• according to Commit 1d36b36, the files path was changed by @xxiong2021 in #1343
• fix: ignore : images by @benoitf in #1339
• Update ramalama-cuda.7.md by @schuellerf in #1344
• Only execute this if /run/ramalama exists by @ericcurtin in #1347

New Contributors

• @arburka made their first contribution in #1319
• @xxiong2021 made their first contribution in #1343
• @schuellerf made their first contribution in #1344

Full Changelog: v0.8.1...v0.8.2